dating painting events through by-products of ochre ...€¦ · article dating painting events...
TRANSCRIPT
Full Terms & Conditions of access and use can be found athttps://www.tandfonline.com/action/journalInformation?journalCode=raaa20
Australian Archaeology
ISSN: 0312-2417 (Print) 2470-0363 (Online) Journal homepage: https://www.tandfonline.com/loi/raaa20
Dating painting events through by-productsof ochre processing: Borologa 1 Rockshelter,Kimberley, Australia
Bruno David, Jean-Jacques Delannoy, Fiona Petchey, Robert Gunn, JillianHuntley, Peter Veth, Kim Genuite, Robert J. Skelly, Jerome Mialanes, SamHarper, Sven Ouzman, Balanggarra Aboriginal Corporation, Pauline Heaney& Vanessa Wong
To cite this article: Bruno David, Jean-Jacques Delannoy, Fiona Petchey, Robert Gunn, JillianHuntley, Peter Veth, Kim Genuite, Robert J. Skelly, Jerome Mialanes, Sam Harper, Sven Ouzman,Balanggarra Aboriginal Corporation, Pauline Heaney & Vanessa Wong (2019): Dating paintingevents through by-products of ochre processing: Borologa 1 Rockshelter, Kimberley, Australia,Australian Archaeology, DOI: 10.1080/03122417.2019.1603263
To link to this article: https://doi.org/10.1080/03122417.2019.1603263
Published online: 17 May 2019.
Submit your article to this journal
View Crossmark data
ARTICLE
Dating painting events through by-products of ochre processing:Borologa 1 Rockshelter, Kimberley, Australia
Bruno Davida,b , Jean-Jacques Delannoyb,c, Fiona Petcheyb,d, Robert Gunna, Jillian Huntleye,Peter Vethb,f , Kim Genuitec, Robert J. Skellya, Jerome Mialanesa,b , Sam Harperf, Sven Ouzmanf,Balanggarra Aboriginal Corporationg, Pauline Heaneyh and Vanessa Wongi
aMonash Indigenous Studies Centre, Monash University, Clayton, Australia; bAustralian Research Council Centre of Excellence forAustralian Biodiversity and Heritage; cLaboratoire EDYTEM, Universit�e Savoie Mont Blanc, Le Bourget du Lac Cedex, France;dRadiocarbon Dating Laboratory, University of Waikato, Hamilton, New Zealand; ePlace, Evolution and Rock Art Heritage Unit,Griffith Centre for Social and Cultural Research, Griffith University, Gold Coast, Australia; fArchaeology and the Centre for RockArt ResearchþManagement, M257, School of Social Sciences, Faculty of Arts, Business, Law and Education, The University ofWestern Australia, Crawley, Australia; gBalanggarra Aboriginal Corporation, Wyndham, Australia; hLettuce Create, Strathpine,Australia; iSchool of Earth, Atmosphere and Environment, Monash University, Clayton, Australia
ABSTRACTThe ‘direct’ dating of rock art has proliferated since the development of accelerator massspectrometry radiocarbon, uranium-series and optically stimulated luminescence dating, yetstill, most rock art is not directly datable due to the mineral nature of the constituent pig-ments. Here we present another method: the recovery and dating by stratigraphic associ-ation of small buried fragments of ochre and dried paint drops deposited onto softsediment surfaces as by-products of paint production and use. These finds also give addedcontextual occupational information for archaeology of painting events. The case is madethrough the example of Borologa 1, a richly decorated Wanjina rockshelter in the Kimberleyregion of northwestern Australia that contains buried hearths, grindstones, earth pigmentsand small fallen spalls of rock containing traces of pigment and paint drops. Results fromexcavation indicate the beginning of Wanjina motifs and associated painting conventions onArt Panel B1 sometime between 2,080–1,160 cal BP and their proliferation in thepast millennium.
ARTICLE HISTORYReceived 29 August 2018Accepted 2 April 2019
KEYWORDSAboriginal sites;archaeological excavationmethods; dating rock art;Kimberley; ochreprocessing; Wanjina
Introduction
The late 1980s and early 1990s saw a notable shiftin the treatment of rock art by archaeologists world-wide. Advances in accelerator mass spectrometry(AMS) radiocarbon dating (capable of dating tinysamples measured in milligrams of carbon), uran-ium-series (U-series) and optically stimulated lumi-nescence (OSL) dating of mineral accretions over orunder rock art, and refined micro-geomorphologicalanalyses of rock surfaces containing paintings andengravings, enabled improved absolute and relativechronologies which could situate rock art more reli-ably in conventional archaeological narratives of thepast (e.g. Watchman 1987, 2000; Watchman et al.2000). These innovations came concurrently with arenewed interest in symbolic expressions and cogni-tive archaeology (e.g. Chippindale and Tacon 1998).What has not been as methodologically forthcom-ing, however, are developments (and improvements)in the logic, planning and methods of excavation tounderstand better the antiquity of artistic activities
within sites, and of artworks on rock surfaces (butsee McNiven et al. 2009). Here we present excava-tion results from Borologa 1, a painted rockshelterin the Kimberley region of northwestern Australia,with an eye to studying the antiquity of iconicWanjina-style rock paintings and their associateddesign conventions. We use ‘Wanjina’ rather than‘Wandjina’ or ‘Wondjina’ as this is the term pre-ferred by local Aboriginal communities in theEast Kimberley, where this work was conducted(see also Akerman 2016). Our aim is to develop anew focus on the recovery of evidence relating toochre processing and use as a means of clarifyingthe antiquity and activities of individual artisticevents – an under-appreciated scale of analysis inarchaeology.
Wanjina
The Kimberley region of northwest Australia con-tains thousands of rock art sites under overhangs,
CONTACT Bruno David [email protected] Monash Indigenous Studies Centre, 20 Chancellors Walk, Monash University, ClaytonCampus, Victoria 3800, Australia� 2019 Australian Archaeological Association
AUSTRALIAN ARCHAEOLOGYhttps://doi.org/10.1080/03122417.2019.1603263
on rockshelter walls, and engraved in the open.There are also related symbolic expressions such asstone arrangements, but we here restrict ourselvesto rock paintings. A number of style sequenceshave been devised to order chronologically therock paintings of the Kimberley (for a summary,see Table 1); we follow Veth et al. (2018) andnote the problematics inherent in the word ‘style’,which are beyond the scope of this paper. Oneof the most iconic styles – Wanjina – consistsmainly of human-like faces sometimes with partialor whole bodies, and often associated with the ser-pent ungud as well as with other motifs suchas animals and plants. These motifs are oftenpainted on prepared white pigment backgrounds.Wanjina rock art continued to be painted intothe twentieth century and, in some areas, to thepresent day (see Morwood et al. 2010). While weknow that Wanjina paintings occur across anarea c. 600 km long (north-south) by c. 700 kmwide (east–west) (for a map, see Akerman 2016),we do not yet know exactly when the style orits various painting conventions began, nor pre-cisely how its commencement interphases withearlier styles. Understanding the antiquity of thestyle has considerable importance for understand-ing the foundations of historically documentedAboriginal world views relating toWanjina imagery.
Any discussion of human occupation and sym-bolic inscription of Kimberley cultural landscapesmust accord primacy to local Aboriginal concep-tions of the Dreaming (called lalai and lalan in thenorthwest and central Kimberley, respectively) andongoing connections to land and cynosuric places(focal places of signification) (e.g. BalanggarraAboriginal Corporation and Kimberley LandCouncil 2011). In a recent volume on the northwestKimberley, Blundell et al. (2017:44) noted the ubi-quitous presence of Wanjina with whom peopleconstantly interact: ‘The country is perceived as theembodied consequence of the actions of powerfulancestral creator beings rather than as a landscapethat has been altered through the actions of geo-logical or climatic events, or by human agents’. In avery real sense, rock art is understood to be a prod-uct of both ancestral and contemporary actionsthat include the ritual attendance (including throughpainting and repainting) of ancestral beingscalled Wanjina.
Wanjina are found exclusively in the Kimberley,mainly on rock surfaces across Worrora,Ngaranyin, Wunambal and Balanggarra lands.Historical and other accounts connect individualWanjina paintings with the locations where theancestral Wanjina spirit-beings lay down in theDreaming (for ethnographic and recent textualaccounts of Wanjina, see Capell 1972; Crawford
Table 1. Style sequence nomenclature.
Proposedsequence
Period
Summary characteristicsWelch 1993 Walsh 2000 Donaldson 2012 Veth et al. 2018
Earliest Archaic Archaic Pecked cupules and abraded groovesIrregular
Infill AnimalIrregular
Infill AnimalIrregular
Infill AnimalStencils, prints, large outline fauna and
anthropomorphs with stippled irregu-lar infill
Bradshaw GroupTasselled Figure Tasselled
BradshawNgunuru Gwion Gwion (tassel) Graceful, finely painted, slim-
bodied anthropomorphs with tasselornamentation, elongated head-dresses and sometimes holding boo-merangs or wearing dillybags. Smallanimals associated
Bent Knee Figure Sash Bradshaw Yowna Gwion Gwion (sash) Elongated anthropomorphs with verytall headdresses, sash-like aprons andflexed knees (dancing?)
KimberleyDynamic Figure
ElegantAction Figures
Dynamic Gwion Gwion (dynamic) Elongated anthropomorphs with littlebody ornamentation and mostlydepicted in active poses(e.g. running)
Straight Part Figure Clothes Peg Figure Dalal Gwion StaticPolychromeFigure
Anthropomorphs in static frontal poseand with conical headdresses. Oftenbichrome (although the more fugi-tive colours may be lost). Barbedspears, hooked spear-throwers andboomerangs present
Painted Hand Clawed Hand Painted Hands Painted Hand Broad-brush outline paintings with seg-mented or grid infill. Individual out-lined hands with tapering fingers
Most recent Wanjina Wanjina Wanjina Wanjina, Contact Monochrome and polychrome Wanjinaanthropomorphs (head, headþbody,or headdress alone) and animals
Contact Contact Anthropomorphs and objects from theEuropean-contact period
2 B. DAVID ETAL.
1968; Elkin 1930; Lommel 1952; Mowaljarlai andMalnic 1993; Schulz 1956; see also Akerman 2016;O’Connor et al. 2008). Repainting of Wanjinaimagery is a ritual action undertaken by initiatedAboriginal men of the appropriate affiliation toCountry and as mediators of the Wanjina them-selves (Blundell et al. 2017:70). These are formalactivities that ensure the regeneration of the land-scape’s fertility and life processes (Blundell et al.2017:70; see also Veth et al. 2018). The knowledgeand connections to the power associated with theWanjina and the life forces they emit make the act ofrepainting sacred (Crawford 1968).
Wanjina originated either from the sky or sea,and travelled, fought and created as they jour-neyed, metamorphosising onto the rock as paint-ings imbued with affective life-forces. Theycontrol fertility, growth, landscape processes, andprovide instruction on social behaviour and howto look after Country (‘Law’), including the rulesrelating to appropriate marriage partners, thelocation of clan estates and the undertaking of rit-uals. People can communicate with Wanjinathrough dreams and can be given songs and dan-ces as well as advice on matters such as paintingand repainting. Wanjina rock images are clearlynot just paintings in local Aboriginal ontologiesbut are themselves sentient beings (e.g. Blundellet al. 2017).
How old is the Wanjina style of painting?
Over the past three decades, radiocarbon and lumi-nescence dating have been applied to determine theage of Wanjina paintings (e.g. Morwood et al. 2010;Roberts et al. 1997; Ross et al. 2016). The customaryrepainting of Wanjina on rock surfaces means thatfor each motif there will probably be a sequence ofpainting events, from first iteration through to themost recent additions. Morwood et al. (2010)obtained a radiocarbon age of 3,780 ± 60 BP (cali-brated at 95.4% probability to 4404–3981 cal BP) onbeeswax that made up a Wanjina head motif, givinga minimum age for the start of Wanjina motifs asan image type. More recent radiocarbon ages werealso obtained on other Wanjina head motifs and onfauna painted in the Wanjina style, indicating thecontinuity of painting events into the colonialperiod: ‘heavily repainted examples only date to thelast 375 ± 35 years – the age of the lowermost ofthirty-eight layers of paint in a Wandjina-style snakedepiction’ (Morwood et al. 2010:6; see alsoWatchman 1987).
Roberts et al. (1997) applied luminescence datingto fossilised mud-dauber wasp (Sceliphron laetum)nests overlying rock art, including an ‘early phase’
Wanjina head from Borologa 1 (also known asWandjina Rock) along the Drysdale River. Threewasp nests (DR1, DR2 and DR6) were dated fromBorologa 1, the one on the Wanjina head (DR6)being sampled sequentially down the height of thenest to test for modern, contaminating sunlightexposure. Two generations of nest building datingto 110 ± 20 and 610 ± 50 years ago were determinedfrom the OSL ages on DR6. Because the nest wasbuilt over the Wanjina painting, the painting itselfmust be at least c. 600 years old. The other two nests(DR1 and DR2) both date to the past 100–150 years(see also Aubert 2012).
The oldest age published for a purportedWanjina-style motif is a minimum OSL age of5,100 ± 240 years for a mud-dauber wasp nestoverlying a striped macropod painting (Ross et al.2016, sample LM-13). The motif is a kangaroowith body and limbs divided into sections withthe entire motif infilled with fine longitudinallines, and the addition of white dotting on theshoulders, forearms and kidneys. Ross et al.(2016:13) suggested that the backbone and kidneysare similar to X-ray art of Arnhem Land to theeast, which has a dated beginning from the mid-Holocene. Ross et al. (2016:26) argued that ‘simpleX-ray elements have been recorded in threeGwion Period figures … but are more commonin the Painted Hand Period motifs where animalsare sometimes depicted with backbone and ribs’.This is of interest as it suggests that the X-raydesign found in the LM-13 motif may be an early,transitional convention that pre-dates the Wanjinastyle itself.
On the current ages available, the ‘proliferationphase’ for classic Wanjina iconography dates to thepast millennium (Morwood et al. 2010). While someearlier design elements that overlap with those ofWanjina style motifs may be several thousand yearsolder, the age of the beginning of most design ele-ments remains unknown.
Borologa 1
Borologa 1 is a 7m high rock stack at the base ofa steep hill on the banks of the Drysdale River.Although Balanggarra Elders of the late 1980s andearly 1990s identified the site as ‘Borologa’ (asrecorded by rock art enthusiast Joc Schmiechen,see Welch 2015:69), it has been popularly knownby visitors to the Drysdale River National Park as‘Wandjina Rock’, after its many Wanjina paintings.During the early dry season when flood watershave abated, the site is located 120m north of theDrysdale River in the northern Kimberley region(Figures 1 and 2). The rock stack consists of
AUSTRALIAN ARCHAEOLOGY 3
south-southwest-tilted quartzite and sandstonelayers (designated rock strata D3–D25) sitting on asubhorizontal bedrock pedestal also comprised ofsuperimposed hard quartzite layers (strataD0–D2), with rock stratum D0 only partiallyexposed above ground (see Figure 3 for a visualrepresentation of the rock strata that make up thestack). Two alcoves at ground level were createdby a combination of geological weathering (rockfall) and human removal of rock strata (the evi-
dence for human modification of the rock will bepublished elsewhere), resulting in sheltered areasunder moderate overhangs with extensive evidenceof bedrock flaking by people, production of rockart, and surface and buried cultural deposits(Figure 3). The soft sediments fronting the southside of the site consist of aeolian sands over anancient alluvial terrace. Today the floor level lies7m above the seasonal high flood level of theDrysdale River.
Kalumburu
Kununurra
Theda Station
Drysdale River
Barking Owl
Camp
plateau
plateau
Borologa
Dry
sdal
e Riv
er
white pigmentsource
100500
km
10005000
m
Wyndham
WesternAustralia
Kimberley
Figure 1. Location of Borologa 1, northwestern Australia (photos: Jean-Jacques Delannoy).
4 B. DAVID ETAL.
The rock art of Borologa 1
Methods
We began by recording the rock art of Borologa 1through detailed photography, drawings and note-taking, followed by digitally enhancing the photos toclarify painting conventions and determining pat-terns of superimposition. Individual paintings wereattributed to established Kimberley rock art styles,as determined from the literature (see below).However, until more chronological details are forth-coming, we do not assume an inviolate unilineartemporal sequence for the style phases or sets ofgraphic conventions, nor do we exclude the possibil-ity of additional ‘styles’.
Results
Borologa 1 is extensively decorated with rock art,most prominently a ceiling panel featuring Wanjinamotifs sheltered under a well-protected alcove in thesoutheastern side of the site (Art Panel B1, seeFigure 3). Small grinding patches are found on hori-zontal rock surfaces in front of all the art panels.Those grinding patches are sometimes on movablerocks, sometimes on the outcropping flat bedrock,either at ground level or slightly more elevated.Grinding patches on the upper surfaces of twoboulders in the alcove that houses Art Panel B1bear traces of red pigment. Other shallow grindingpatches are positioned near painted rock art panels(Figure 4) around the stack, suggesting that ochre
Figure 2. (A) Borologa 1 in its local setting. (B) Borologa 1 (photos: Jean-Jacques Delannoy).
AUSTRALIAN ARCHAEOLOGY 5
was typically processed on rock palettes close towhere the resulting paint would be used.
The rock art is found in 40 distinct art panelsthat together cover much of Borologa 1’s western,southern and eastern sides (the rough, weathered
rock on the northern side is exposed to the ele-ments and shows no sign of artworks) (Figure 3).The art panels are found along 26m of rock sur-face, spanning a height of some 2m from justabove the present ground level, the combined art
Figure 3. Plan (top) and side view (bottom) of Borologa 1 showing the monolith’s rock strata and location of excavationsquares (artwork: Jean-Jacques Delannoy).
6 B. DAVID ETAL.
area totalling 50.4m2. A preliminary countrecorded 446 motifs in a wide range of styles, col-ours, and states of preservation. Following themost recently established Kimberley style catego-ries (after Donaldson 2012; Ouzman et al. 2017;Veth et al. 2018; Walsh 2000) (see Table 1), rockart from the early Irregular Infill Animal, TasselledGwion, Dynamic Gwion, and Static PolychromeFigure styles, and the more recent Painted Handand Wanjina styles is present (Figure 5). Hand
stencils, hand-prints and feather prints are alsopresent. A number of distinctive stout anthropo-morphic figures with vertical stripe infill alsooccur on several panels. Welch (2015) consideredthem to date probably to the earlier or later stagesof the Painted Hand style, as they have characteris-tics of both bracketing styles (Static PolychromeFigure and Wanjina styles, respectively) (DavidWelch, pers. comm. 2018). Similarly, a distinctcomposition of 31 motifs, involving a type of
Figure 4. Grinding patches on rock surfaces near rock art panels, Borologa 1 (photos: Robert Gunn, Leigh Douglas andBruno David).
AUSTRALIAN ARCHAEOLOGY 7
unclassified stick figure, is most likely from theStatic Polychrome Figure style. Borologa 1 alsocontains a number of other motifs that cannot beallocated to particular styles. It is important tonote that while the relative chronology of some ofthese style phases is undisputed – especially Gwionalways preceding Wanjina paintings – the absoluteantiquity and degree of overlap between manystyles remains uncertain, as does the chronologicalposition of a number of individual motifs (e.g.hand stencils and hand prints) in thebroader schema.
At Borologa 1, only one Gwion style is repre-sented on any given art panel, although paintingsfrom different styles may occur on adjacent panels.Panels with Gwion figures may also contain PaintedHand style animal or Wanjina motifs. The Gwionmotifs with tassel waist bands only occur on panelsat the eastern end of the rock stack, whereasDynamic Gwion and Static Polychrome Figures arerestricted to those on the west. Art Panel E4, thelargest vertical art panel on the western side, contains13 Static Polychrome Figures in a range of colours,forms, associated artefacts and states of preservation.The Static Polychrome Figures are painted in mono-chrome red or yellow, or bichrome red and yellow,
while the Tasselled Gwion and Dynamic Gwionoccur only in monochrome red to mulberry. Most ofthe stout striped figures are in monochrome red (theexceptions being yellow or yellow and red) and occuron panels around the three decorated faces of theboulder, showing no preference for association withGwion or Wanjina images.
Most of the Gwion figures are weathered andincomplete, but most appear to be between 10 cmand 50 cm tall based on comparisons with com-plete figures at other sites. There is little variationin height between the different types of Gwion fig-ures. The stout striped figures are of similar sizeto the Gwion, whereas the two Painted Handmotifs are larger, being 71 cm and 165 cm longrespectively. The Wanjina vary considerably insize, from 18 cm to 119 cm, the variation largelydue to whether a head or head and torso isdepicted, and also to the size of the art panel onwhich it is placed as they are proportioned to fillthe panel.
The Wanjina, whether represented by a singularhead, upper body, or simply by an arc headdress,occur both in red monochrome and in polychrome(a white base decorated with red and/or black, yel-low or cream). No Wanjina is represented with acomplete body at Borologa 1, although such full-bodied examples are known from elsewhere in theKimberley, the closest being c. 40 km to the southand to the west.
Art Panel B1
Art Panel B1 is located on the ceiling of the south-eastern alcove (Figure 6), adjacent to the excava-tion squares reported below. The art consists of107 paintings, two dry pigment drawings and twohand stencils. Art Panel B1 is the most denselydecorated panel at the site, and is dominated bylarge and impressive Wanjina paintings, some withassociated Wanjina-style animals. The Wanjina,whether as singular head or as head and upperbody, occur both in monochrome red and poly-chrome (up to four colours). Polychrome Wanjinaare limited to the large ceiling of the easternalcove, while monochrome Wanjina occur both onthis ceiling panel and on other, smaller, verticalwall panels on the eastern and southern sides ofthe site. No Wanjina occur on art panels on thewestern side of the site where Static PolychromeFigure motifs predominate.
Some of the polychrome Wanjina are paintedon a white background, others not. All bar one ofthe polychrome Wanjina-with-torso show evidenceof over-painting in the form of another Wanjinasuperimposed over an earlier one. These superim-posed Wanjina have the same orientation as the
Figure 5. The apparently most recent Wanjina face on ArtPanel B1 in bright red over an earlier head and upper bodyWanjina (photo: Robert Gunn).
8 B. DAVID ETAL.
underlying ones but are slightly offset. Onesequence shows at least three superimposedWanjina (Figure 7). Superimposition with non-Wanjina art is, however, more extensive, with atleast eight layers being deciphered from Art PanelB1 (Table 2); Art Panel E4 on the western wall ofthe boulder exhibits five instances of superimpos-ition, and further examples are evident on otherart panels at the site.
Art Panel B1 sits just above the northwesternedge of the excavation squares reported here (seeFigure 8). A large part of the eastern half of the ceil-ing was painted in white to form a background forWanjina figures, while other Wanjina images furtherto the west on Art Panel B1 do not have a pigmentbackground. The white background was repainted atleast once and is the only area of the site with
evidence for such a large quantity of white pigmentbeing employed. The earlier white background,however, overlies another Wanjina figure that lacksa white background (Figure 9), indicating that theuse of the white background for Wanjina figuresdoes not signal the earliest depiction of Wanjina atthis site. Based on patterns of superimposition, theseWanjina are among the most recent images on thispanel. The most recent image of all appears to be asmall monochrome Wanjina head positioned overthe torso of an earlier polychrome Wanjina(Figure 5).
The earliest painting on Art Panel B1 is an out-lined and internally decorated snake-like image onthe outer edge of the panel (Figure 10). This imageconforms to the style of the Painted Hand style.All of these images are from the latest styles in theknown rock art of the Kimberley (Table 1).
Figure 6. Photomosaic of Art Panel B1; the ceiling of the eastern alcove (photo: Robert Gunn).
Figure 7. Three superimposed Wanjina figures (A–C) on ArtPanel B1 (photo: Robert Gunn).
Table 2. Superimposition layers from the area of Art PanelB1 shown in Figure 5.Relative chronology Art layer characteristics
Most recent layer Area of bright red dashesWanjina with black centred eyes, red nose, and
yellow striped red headdressRemnant of an outlined figure in red with legs
together, splayed feet, and short toesRemnant of a dashed infill design fragment
in redWanjina with white face, red eyes and plain
band headdressWanjina with white face and short-rayed
band headdressWhite solid area (background to Wanjina)
Earliest discerniblelayer
Outlined dumbbell shape in black (or discol-oured red)
AUSTRALIAN ARCHAEOLOGY 9
The excavation
Methods
Three pits were excavated, all fronting Art PanelB1, the eastern alcove ceiling densely decoratedwith Wanjina paintings. Here we restrict our dis-cussion to the pit closest to the art panel. The1.2m wide by 1.6m long pit consists of four60� 60 cm squares: C5, C6, D5 and D6; and two
smaller 60� 40 cm squares: DV and DVI. In add-ition, a 20� 20 cm square (Square D7) was exca-vated against the northeastern wall of Square D6after completion of the other six excavations(Figures 8 and 11). Square D7 was dug to compen-sate for the loss of much of the excavated materialfrom Squares C5 and C6 in a helicopter mishaplate in the final field season.
Figure 8. Excavation pits relative to Art Panel B1, excavations in progress, with Squares C5 and C6 (those closest to theWanjina ceiling) not yet begun. The excavations reported in this paper concern the front pit. Pits E4 and F5-G5 are shown onthe edge of and outside the dripline at rear of photo (photo: Bruno David).
10 B. DAVID ETAL.
Each Square was typically excavated in 1–2 cmthick arbitrary excavation units (XUs) followingthe stratigraphy of cultural layers (stratigraphicunits SU1-SU5). However, XUs were often 3–4 cmthick in the more rocky, underlying culturally
sterile horizons (SU6-SU7). Artefacts �2 cm-long,and selected smaller items including charcoal, wereindividually collected and plotted onto the XUrecording forms (Table 3). The excavated sedi-ments from the main pit were sieved in the fieldin 2mm-mesh, with all retained materials subse-quently sorted at the Monash University archae-ology laboratories. The Square D7 deposits werebulk-bagged by XU in the field, then sieved andsorted in the laboratory. Small, bulk sediment sam-ples from each XU of each square were kept forsediment analyses. The lower levels of eachsquare reached culturally sterile, impenetrablemassive roof-fall rock in SU6 or SU7. It isnot known whether further cultural layerslie beneath.
Results
Stratigraphy and radiocarbon dating
Table 4 describes the seven SUs identified in theexcavation (Figures 12 and 13). Thirty-eight AMSradiocarbon ages were obtained on individual
Figure 9. Earlier Wanjina figure (A) overpainted by whitearea background (B) at point (C). The pale green colour at(D) is a later white repainting over area (B), while (E) is partof the more recent Wanjina figure painted on to background(D) (enhancement in DStretch_ywe10) (photo and enhance-ment: Robert Gunn).
Figure 10. Parts of Art Panel B1. (A) Motif with outline, form and aspects of infill in some respects consistent with earlyWanjina, but also sharing attributes of the Irregular Infill Animal style, overlying (B) an internally decorated snake-like imagethat conforms to the style of the Painted Hand style (the lower photo is enhanced with DStretch_lrd10). The apparent inver-sion of some stylistic conventions highlights one of the problems with the currently proposed stylistic sequences forKimberley rock art (photos and enhancement: Robert Gunn).
AUSTRALIAN ARCHAEOLOGY 11
pieces of charcoal. Twenty-nine were recorded insitu during excavation, the other nine came fromthe sieves (Table 5). To determine the age ofonset, end and duration of each phase of sedimentdeposition, we have used OxCal v4.3 (BronkRamsey 2018) to develop a Bayesian SequenceAnalysis whereby radiocarbon ages are con-strained by the stratigraphic structure of thedeposit (Bronk Ramsey 2009). We note that nei-ther the unmodelled ages nor the BayesianSequence Analysis takes into account potentialinbuilt ages in the wood charcoal (e.g. Allen &Wallace 2007; Schiffer 1987), which in this envir-onment could be up to a few decades old, or moreif the wood is cypress pine (Callitris intratropica).Wk-44512 at the base of SU1B was manuallyrejected prior to Bayesian Sequence Analysis as itcontains bomb carbon indicating a modern (post-AD 1950) age. It is the only radiocarbon age witha post-bomb result and is associated with near-surface charcoal from a recent bushfire. Theremaining samples were grouped into sevenchronostratigraphic Phases, from top to bottom(Figure 15):
� Base of SU1B (includes SU1B-SU2 interface):450–120 to 270–190 cal BP. This Phase relates towhen the very top of SU2 was the ground surface.
� SU2: 480–340 to 450–120 cal BP.� SU3 hearth layer (includes SU3-SU4 interface):
630–500 to 480–340 cal BP. It includes a num-ber of stratigraphically inter-connected hearths
and a plant-processing basal grindstoneand topstone.
� Base of SU4 (includes SU4-SU5 interface), whenthe very top of SU5 was the ground surface:1,110–970 to 910–760 cal BP.
� SU5: This consists of the bulk of the SU5 deposit,stratigraphically above the lower two Phases listedbelow: 2,080–1,610 to 1,600–1,160 cal BP.
� Base of SU5 (includes SU5-SU6 interface), whenthe fallen rock at the top of SU6 was the groundsurface: 2,370–2,180 to 2,320–2,110 cal BP.
� Hearth at base of SU5: 3,000–2,500 to2,690–2,270 cal BP. This small hearth is locatedon the rock surface at the junction of SquaresC5, C6, D5 and D6, at the very base of SU5.
The earliest five Phases were separated by uni-form, sequential age boundaries representing theage hiatus identified between events. The base ofSU1B, SU2 and the SU3 hearth layer Phases arechronostratigraphically differentiated and weretherefore separated by contiguous age boundariesin the model (see Bronk Ramsey 2009). We testedthe internal consistency of the calibrated agesusing an outlier detection method that enables aprobabilistic measure of the degree to which samplesappear to be outliers and then calculates an offsetrelative to the broader chronostratigraphic contextwithin which it is found (Bronk Ramsey et al. 2010).In this instance, a General t-type Outlier Model wasinset into the Bayesian Sequence Analysis and allages were assigned a prior outlier probability of 0.05.
Figure 11. Borologa 1, showing location of excavation squares relative to Art Panel B1 (photo: Bruno David).
12 B. DAVID ETAL.
Table3.
Detailsof
XUs.
XUSU
MeanDepth
atTop(cm)
MeanDepth
atCentre
(cm)
MeanDepth
atBase
(cm)
Mean
Thickness(cm)
Area
(m2 )
Weigh
t(kg)
Volume(litres)
Roots(g)
Leaves
(g)
Seeds(g)
Non
-Artefactual
Rock
�2mm
wide(g)
Square
D5
11A
01.3
2.5
2.5
0.36
12.5
9.0
13.5
0.9
0.1
283.5
21A
þ1B
2.5
3.8
5.1
2.6
0.36
12.4
9.0
5.0
<0.1
18.9
31B
þ25.1
6.1
7.0
1.9
0.36
12.8
9.0
9.3
5.2
41Bþ2
7.0
8.4
9.8
2.8
0.36
13.0
9.0
4.9
0.1
1.5
52
9.8
10.8
11.7
1.9
0.36
7.6
5.0
0.4
0.5
62
11.7
13.7
15.6
3.9
0.36
24.0
19.5
7.5
<0.1
72
15.6
17.1
18.6
3.0
0.36
18.4
14.0
4.9
0.3
0.5
82
18.6
20.4
22.1
3.5
0.36
14.6
12.0
3.2
0.6
92þ
322.1
23.3
24.5
2.4
0.36
8.0
6.5
3.0
1.2
102þ
324.5
25.4
26.3
1.8
0.36
n/a
7.0
4.1
13.3
112þ
326.3
27.5
28.7
2.4
0.36
12.6
10.0
14.7
113.2
122þ
328.7
30.0
31.3
2.6
0.33
15.2
12.0
9.9
95.3
13a
2þ3þ
431.3
32.9
34.5
3.2
0.20
4.8
4.0
0.1
7.8
13b
2þ3
31.3
32.9
34.5
3.2
0.13
7.0
6.0
2.2
20.3
14a
3þ4
34.5
35.6
36.7
2.2
0.23
6.4
4.5
2.6
3.6
14b
3þ4
34.5
35.6
36.7
2.2
0.10
7.0
4.0
1.6
2,673.6
15a
3þ4
36.7
38.1
39.5
2.8
0.26
10.4
8.0
1.6
9.3
15b
336.7
38.1
39.5
2.8
0.07
2.8
2.5
1.3
5.0
16a
439.5
40.7
41.9
2.4
0.30
7.8
7.0
0.5
24.1
16b
339.5
40.7
41.9
2.4
0.03
2.4
2.0
0.4
1.9
17a
441.9
43.2
44.5
2.6
0.32
7.6
7.0
0.1
343.4
17b
3þ4
41.9
43.2
44.5
2.6
0.02
3.4
3.0
0.8
13.3
184þ
544.5
46.3
48.1
3.6
0.34
17.2
12.5
2.5
2,071.1
194þ
548.1
50.1
52.1
4.0
0.34
17.0
12.5
205
52.1
54.5
56.9
4.8
0.34
24.5
18.0
1,082.5
215þ
656.9
58.8
60.6
3.7
0.34
22.6
15.0
0.8
9,187.7
225þ
660.6
63.4
66.1
5.5
0.34
39.8
26.0
21,800.0
236
66.1
69.4
72.6
6.5
0.07
9.5
7.0
0.6
7,273.8
246
72.6
74.2
75.7
3.1
0.07
7.0
5.0
0.4
5,100.9
256
75.7
79.3
82.9
7.2
0.34
48.0
35.0
1.1
40,296.9
266
82.9
86.1
89.3
6.4
0.34
36.4
24.0
0.7
29,200.0
276
89.3
91.7
94.1
4.8
0.34
20.0
12.5
0.9
15,800.4
Sub-total
3.5±1.5
452.7
337.5
98.6*
0.9
0.5
135,449.3*
Square
D6
11A
00.7
1.4
1.4
0.36
8.0
6.0
15.7
1.5
0.3
72.1
21B
1.4
2.7
3.9
2.5
0.36
14.2
10.5
8.5
<0.1
64.0
31B
þ23.9
5.1
6.3
2.4
0.36
11.0
7.5
8.9
44.0
41Bþ2
6.3
7.4
8.5
2.2
0.36
12.0
8.5
30.0
17.3
52
8.5
9.5
10.5
2.0
0.36
10.0
7.0
29.0
0.1
1.4
62
10.5
12.4
14.2
3.7
0.36
21.4
16.5
12.1
<0.1
2.9
(continued)
AUSTRALIAN ARCHAEOLOGY 13
Table3.
Continued.
XUSU
MeanDepth
atTop(cm)
MeanDepth
atCentre
(cm)
MeanDepth
atBase
(cm)
Mean
Thickness(cm)
Area
(m2 )
Weigh
t(kg)
Volume(litres)
Roots(g)
Leaves
(g)
Seeds(g)
Non
-Artefactual
Rock
�2mm
wide(g)
72
14.2
16.1
17.9
3.7
0.36
20.4
15.0
8.5
2.6
82þ
317.9
19.3
20.6
2.7
0.36
10.2
8.0
87.8
0.2
3.5
92þ
320.6
21.6
22.5
1.9
0.36
7.6
6.0
59.1
<0.1
11.6
10a
2þ3
22.5
23.5
24.4
1.9
0.17
n/a
n/a
3.0
128.7
10b
2þ3
22.5
23.5
24.4
1.9
0.19
n/a
n/a
8.0
69.3
11a
2þ3
24.4
25.6
26.8
2.4
0.17
n/a
n/a
3.5
14.4
11b
324.4
25.6
26.8
2.4
0.19
6.0
5.0
4.5
3,176.0
12a
2þ3þ
426.8
28.4
29.9
3.1
0.22
8.6
7.5
1.6
5.0
12b
3þ4
26.8
28.4
29.9
3.1
0.09
6.4
5.0
4.0
17.9
13a
3þ4
29.9
31.7
33.5
3.6
0.15
2.8
2.0
0.1
0.7
13b
3þ4
29.9
31.7
33.5
3.6
0.15
6.6
5.0
5.7
3.9
143þ
433.5
34.6
35.7
2.2
0.05
1.2
1.0
0.3
0.4
154
35.7
37.4
39.0
3.3
0.05
2.4
2.0
0.9
7.9
164
39.0
40.2
41.4
2.4
0.05
1.8
1.0
0.4
29.1
174þ
541.4
42.6
43.8
2.4
0.05
1.6
1.0
1.0
83.4
184þ
543.8
46.0
48.1
4.3
0.05
2.0
1.5
0.1
0.1
123.8
265
48.1
49.3
50.5
2.4
0.06
3.8
2.5
275þ
650.5
52.5
54.4
3.9
0.30
10.8
7.5
285þ
654.4
55.9
57.3
2.9
0.06
7.3
5.0
295þ
657.3
59.2
61.1
3.8
0.36
22.3
14.5
0.8
9,306.3
305þ
661.1
63.6
66.0
4.9
0.36
32.2
21.5
1.1
24,102.7
316
66.0
68.7
71.4
5.4
0.15
11.7
9.0
0.8
7,536.9
326
71.4
73.3
75.0
3.6
0.15
11.5
7.5
0.2
8,154.6
336
75.0
78.7
82.3
7.3
0.36
43.6
30.5
1.0
34,774.7
346
82.3
85.3
88.2
5.9
0.36
38.5
26.0
2.6
29,200.4
356þ
788.2
91.5
94.8
6.6
0.36
42.7
29.5
0.6
34,700.0
366þ
794.8
97.6
100.4
5.6
0.06
7.8
5.5
0.8
5,800.0
377
100.4
104.7
109.0
8.6
0.36
22.1
14.5
0.8
19,200.0
Sub-total
3.6±1.8
408.5
289.5
301.4*
1.5
0.7
176,655.5*
Square
DV
11A
þ1B
01.5
2.9
2.9
0.24
13.8
9.0
17.8
24.6
0.2
1,950.4
21B
þ22.9
4.9
6.9
4.0
0.24
10.6
7.5
28.0
34.6
31Bþ2
6.9
7.8
8.6
1.7
0.24
6.4
4.5
31.0
41.4
42
8.6
10.1
11.5
2.9
0.24
10.0
7.0
35.4
10.5
52
11.5
13.6
15.7
4.2
0.24
13.4
10.0
10.1
0.9
62
15.7
17.2
18.7
3.0
0.24
9.8
7.5
7.3
379.8
72
18.7
20.5
21.2
3.5
0.24
11.6
10.0
3.5
0.7
82
21.2
22.6
23.9
2.7
0.24
8.8
8.0
44.6
0.4
92
23.9
25.3
26.7
2.8
0.24
7.6
7.0
27.8
0.4
102
26.7
27.5
28.3
1.6
0.24
6.0
5.0
7.7
1.1
112
28.3
29.7
31.0
2.7
0.24
7.4
6.8
5.1
1.2
(continued)
14 B. DAVID ETAL.
Table3.
Continued.
XUSU
MeanDepth
atTop(cm)
MeanDepth
atCentre
(cm)
MeanDepth
atBase
(cm)
Mean
Thickness(cm)
Area
(m2 )
Weigh
t(kg)
Volume(litres)
Roots(g)
Leaves
(g)
Seeds(g)
Non
-Artefactual
Rock
�2mm
wide(g)
122
31.0
32.2
33.3
2.3
0.24
7.2
6.0
4.2
3.7
132
33.3
34.5
35.7
2.4
0.24
7.0
6.0
6.7
9.5
142þ
335.7
37.2
38.6
2.9
0.24
8.8
8.0
3.9
234.7
152þ
3þ4
38.6
40.6
42.5
3.9
0.24
12.4
11.0
3.0
2,153.7
162þ
3þ4
42.5
43.4
44.3
1.8
0.24
6.0
5.0
1.4
28.3
17a
3þ4
44.3
45.7
47.1
2.8
0.08
2.4
1.5
0.5
2.9
17b
344.3
45.7
47.1
2.8
0.16
6.4
5.0
1.4
2,181.0
18a
3þ4
47.1
48.5
49.9
2.8
0.11
2.2
1.5
2.3
721.0
18b
347.1
48.5
49.9
2.8
0.13
8.8
7.0
1.8
2,186.0
19aþ
b31
4þ5
49.9
52.4
54.9
5.0
0.24
14.3
11.0
203þ
4þ5
54.9
56.8
58.7
3.8
0.24
14.1
9.0
215
58.7
60.5
62.2
3.5
0.24
16.4
9.5
0.8
8,871.0
225þ
662.2
64.6
67.0
4.8
0.24
19.3
13.0
1.2
10,167.8
235þ
667.0
70.0
72.9
5.9
0.24
25.4
16.5
0.8
15,381.3
246
72.9
75.3
77.6
4.7
0.24
22.2
14.5
0.1
13,573.2
256
77.6
79.8
81.9
4.3
0.24
19.1
14.0
0.6
13,400.0
266
81.9
84.2
86.5
4.6
0.24
20.3
14.0
0.2
15,800.0
276
86.5
89.0
91.4
4.9
0.24
18.4
13.0
0.7
15,200.0
286
91.4
94.9
98.3
6.9
0.24
29.3
19.0
2.3
21,200.0
296
98.3
99.6
100.9
2.6
0.24
13.3
9.0
0.8
10,100.0
Sub-total
3.5±1.3
378.7
275.8
251.0*
24.6
0.2
133,635.5*
Square
DVI
11A
þ1B
01.0
1.9
1.9
0.24
8.8
6.5
43.5
2.0
0.1
22.6
21B
1.9
4.0
6.0
4.1
0.24
12.8
9.0
15.2
0.1
81.9
31B
þ26.0
7.2
8.3
2.3
0.24
7.4
5.0
5.9
39.4
41Bþ2
8.3
9.9
11.5
3.2
0.24
9.0
7.0
8.2
0.1
16.7
52
11.5
13.7
15.8
4.3
0.24
13.8
10.0
7.7
0.1
3.4
62
15.8
17.3
18.7
2.9
0.24
9.6
7.5
7.6
0.9
72
18.7
20.1
21.5
2.8
0.24
9.0
7.0
4.5
0.9
82
21.5
22.8
24.0
2.5
0.24
8.6
7.0
36.5
0.6
92þ
324.0
25.2
26.3
2.3
0.24
7.4
6.5
10.3
2.2
10a
2þ3
26.3
27.3
28.2
1.9
0.18
5.2
5.0
6.0
0.4
10b
2þ3
26.3
27.3
28.2
1.9
0.06
1.8
1.5
0.4
5.2
11a
228.2
29.5
30.8
2.6
0.12
4.0
4.0
3.2
0.2
11b
2þ3þ
428.2
29.5
30.8
2.6
0.12
2.5
3.2
1.3
5.0
1221
3þ4
30.8
32.3
33.8
3.0
0.24
9.2
7.5
4.6
344.3
132þ
3þ4
33.8
35.3
36.7
2.9
0.24
8.8
8.0
3.5
37.2
142þ
3þ4
36.7
38.3
39.8
3.1
0.24
9.2
8.0
2.3
125.4
15a
439.8
41.3
42.8
3.0
0.16
4.0
3.5
0.7
1.0
15b
2þ3
39.8
41.3
42.8
3.0
0.08
8.4
6.0
1.3
2,320.8
16a
442.8
43.7
44.5
1.7
0.15
3.3
2.5
0.3
(continued)
AUSTRALIAN ARCHAEOLOGY 15
Table3.
Continued.
XUSU
MeanDepth
atTop(cm)
MeanDepth
atCentre
(cm)
MeanDepth
atBase
(cm)
Mean
Thickness(cm)
Area
(m2 )
Weigh
t(kg)
Volume(litres)
Roots(g)
Leaves
(g)
Seeds(g)
Non
-Artefactual
Rock
�2mm
wide(g)
16b
342.8
43.7
44.5
1.7
0.09
2.8
2.5
0.8
5.9
17a
444.5
46.0
47.4
2.9
0.14
5.4
5.5
0.6
25.1
17b
344.5
46.0
47.4
2.9
0.10
3.8
3.0
1.1
731.1
18a
4þ5
47.4
49.1
50.8
3.4
0.14
n/a
n/a
1,365.6
18b
347.4
49.1
50.8
3.4
0.10
n/a
n/a
0.3
11.4
193þ
4þ5
50.8
52.1
53.3
2.5
0.24
10.0
9.0
20aþ
b3þ
4þ5
53.3
54.7
56.1
2.8
0.24
9.8
6.5
300.0
215
56.1
57.7
59.3
3.2
0.24
12.5
9.0
1,700.0
225þ
659.3
62.1
64.9
5.6
0.24
23.2
14.0
13,200.0
235þ
664.9
77.4
69.9
5.0
0.24
22.9
15.5
11,000.0
246
69.9
72.8
75.6
5.7
0.22
24.1
16.5
0.7
13,547.2
256
75.6
78.3
80.9
5.3
0.21
17.0
10.5
0.5
11,900.2
266
80.9
83.3
85.7
4.8
0.24
21.6
15.5
0.1
15,000.0
276þ
785.7
87.8
89.9
4.2
0.24
18.0
13.5
0.5
14,400.0
287
89.9
92.0
94.0
4.1
0.24
16.2
12.0
0.4
10,900.0
297
94.0
98.1
102.2
8.2
0.24
58.6
40.0
1.0
44,700.0
Sub-total
3.5±1.4
388.7
287.7
169.0*
2.0
0.4
44,700.0*
Square
D7
11b
þ20
0.5
1.0
1.0
0.04
0.6
n/a
0.5
2.4
22
1.0
2.0
3.0
2.0
0.04
1.1
n/a
0.4
0.6
32
3.0
3.9
4.7
1.7
0.04
1.0
n/a
1.0
0.2
42
4.7
5.6
6.4
1.7
0.04
0.9
n/a
0.6
<0.1
52
6.4
7.3
8.1
1.7
0.02
0.8
n/a
0.5
0.1
62
8.1
11.9
15.6
7.5
0.04
5.7
n/a
0.5
0.2
72þ
315.6
17.8
20.0
4.4
0.04
2.2
n/a
1.5
5.3
83
20.0
21.1
22.1
2.1
0.04
1.0
n/a
0.3
22.3
93
22.1
23.1
24.1
2.0
0.04
0.9
n/a
0.3
0.4
103þ
424.1
25.2
26.2
2.1
0.04
1.2
n/a
0.1
1.1
113þ
426.2
27.1
28.0
1.8
0.04
1.0
n/a
0.3
124
28.0
29.2
30.3
2.3
0.04
1.1
n/a
<0.1
0.4
134
30.3
31.4
32.4
2.1
0.04
1.2
n/a
0.1
1.3
144
32.4
33.6
34.8
2.4
0.04
1.3
n/a
0.1
3.1
154þ
534.8
35.9
36.9
2.1
0.04
1.2
n/a
0.1
45.0
164þ
536.9
38.0
39.0
2.1
0.04
1.3
n/a
0.6
162.7
174þ
539.0
40.6
42.1
3.1
0.04
1.8
n/a
0.5
530.9
185
42.1
43.3
44.5
2.4
0.04
1.7
n/a
<0.1
449.2
195
44.5
46.1
47.6
3.1
0.04
1.9
n/a
0.1
830.5
205þ
647.6
49.4
51.2
3.6
0.04
2.0
n/a
0.1
1,240.4
215þ
651.2
52.8
54.3
3.1
0.04
2.3
n/a
0.1
1,534.7
226
54.3
58.3
62.3
8.0
0.04
5.1
n/a
0.1
3,660.0
Sub-total
2.8±1.8
37.3
n/a
7.7
8,491.2
Total
3.3±1.9
2628.1
1882.5�
860.5*
29.0
1.9
853,553.4*
SUnu
mbers
inbo
ldcontrib
utemostto
that
XU’ssediments.Depthsarebelow
the2016–2017grou
ndsurface.
XUshigh
lighted
ingrey
arethoseforwhich
totalweigh
tsof
rootsandno
n-artefactualrocksareno
tavailable,
asthosesieved
materialswerelost
inahelicop
teraccident
(Squ
ares
C5andC6
areom
itted
asmostmaterialswerelost
inthat
accident).
� excludesweigh
tsfrom
missing
XUs(greyrows).
16 B. DAVID ETAL.
Table 4. Description of sediments by SU, Borologa 1 Squares C5-C6-D5-D6-D7-DV-DVI.SU Typical thickness Description pH Dry Munsell
1 SU1A þ SU1B: 5–10 cm This layer is divided into two sub-SUs. SU1A is the currentloose surface sand. It is largely devoid of vegetationgrowth but contains minor leaf litter. The interface withSU1B is fairly marked, taking place over c. 2–3 cm depth.The base of SU1A is uneven as it meets the slightly con-solidated surface of SU1B. SU1B sediments consist ofloamy sands that are similar to the SU1A sands, albeitrootlets c. 1mm in diameter grow in SU1B only.Sediments in both SU1A and SU1B are homogeneous intexture and colour. The interface with SU2 is flat butpoorly defined, taking place over 3–4 cm depth.
SU1A: 4.8 SU1B: 4.7 10YR 5/4 (yellowish brown)
2 15–35 cm Slightly compact and poorly consolidated homogeneousloamy sand. Dry at time of excavation, becoming slightlymore humid with depth. Occasional rootlets up to 3 cmdiameter grow at all angles; at c. 25 cm depth, rootletstend to grow horizontally at the SU2-SU3 interface. Thebase of SU2 is usually well-defined, taking place over avertical distance of 2–4 cm with the appearance of SU3.However, the SU2-SU3 interface is diffuse where the char-coal-rich SU3 is poorly defined.
4.5 7.5YR 5/4 (brown)
3 3–15 cm A charcoal-rich hearth layer. The upper surface of SU3 (as itinterfaces with SU2 above it) is not flat; its base interfaceswith SU4 typically over a depth of 3–4 cm and undulateseven more than its upper surface, signalling connectingcharcoal-rich lenses of varying depths. The well-definedcharcoal lenses signal the presence of three adjacenthearths (one across much of Squares C6 and D6; a secondalong the southeastern side of Squares DV and DVI; athird along the southwestern side of Squares DV and D5onto the southeastern edge of Square C5). Each hearthwas spatially continuous with its nearest neighbour, sothat together they constituted a band of ash and charcoaltypically 3–5 cm-thick. A lack of horizontal and verticalseparation between the hearths – the total absence ofash-and-charcoal-free sand between them – indicatestheir contemporaneity as a single occupational event con-taining multiple, usually small hearths.
4.5 10YR 4/2 (dark grey-ish brown)
4 5–20 cm Slightly more compact loamy sand than in SU1-SU3.Sediments are homogeneous throughout. The base ofSU4 is flat; the interface with SU5 is marked, taking placeover a vertical distance of 1–2 cm. Rootlets c. 1 mm wideare present.
4.6 7.5YR 5/4 (brown)
5 5–15 cm The upper surface of SU5 is reasonably well defined, consist-ing of compact and moderately consolidated ashy, darkyellowish brown loamy sand. The SU5 upper and lowerstratigraphic boundaries are generally horizontal fromsouthwest to northeast, but slant down slightly towardsthe southeast. Texture and colour are homogeneousacross the pit and with depth, but the lower part of SU5appears more ‘chaotic’ in the sense that its base is lessregular as it interfaces with the uneven upper levels ofSU6. The 7–10 cm-thick interface with SU6 is diffuse, pen-etrating into the upper rocks of SU6. At the junction ofSquares C5, C6, D5 and D6, the flat upper surface of therock at the base of SU5 is fire-stained (Figure 14), andcharcoal is relatively abundant, indicating the location ofa hearth.
5.1 10YR 4/4 (dark yellow-ish brown)
6 c. 35 cm A culturally sterile layer of massive rock fall angling downtoward the southeast, but lying horizontally from south-west to northeast. The sandstone rock is friable, havingfractured in situ and during excavation breaking up intypically 5–10 cm-thick plates; it was excavated throughin all squares, without reaching bedrock. Rootlets up to c.1 cm in diameter grow in the interstices between frag-menting rock plates. SU6 represents superimposed largeslabs of rock-fall. The southeast edges of Squares D5 andD6, onto DV and DVI, mark the area beyond the driplineand are thus exposed to rainfall. In these squares, muchof the rock is highly fragmented and weathered.
5.2 5YR 5/4 (reddish brown)
7 Culturally sterile layer. Large rock slab surface sloping downsouthward at the base of the pit. The angle of the slopesuggests that this represents a large collapsed slab ofrock from an originally deeper overhang. The excavationcould not penetrate into this hard rock except in theupper levels of SU7 in Squares DV and DVI (the moreweathered area beyond the dripline exposed to rain).
5.2 5YR 4/4 (dark reddish brown)
AUSTRALIAN ARCHAEOLOGY 17
This allows outliers to be either too young or tooold, and down-weighs their influence in the model.
This model identified two major outliers, Wk-45962 and Wk-45969 with outlier probabilities of100% and 89% respectively. The small charcoalsample of Wk-45962 gave a radiocarbon age of1,938 ± 16 BP, which has been calibrated to1,930–1,820 cal BP, a period not otherwise repre-sented in the excavated squares. It probably camefrom hearth or bushfire charcoal originatingnearby, either within the site or from the broaderlandscape. Wk-45969 came from adjacent to andslightly above the base of a large grindstone inSquare D6 (see below); it dates to 2,259 ± 16 BP,calibrated to 2,350–2,180 cal BP. This period iswell represented in the excavated squares (the‘Base of SU5’ phase), but all the other charcoalsamples of that age came from 5–16 cm deeperthan Wk-45969. This latter charcoal sample isinterpreted to have been scuffed up by a few centi-metres during the positioning of the large grind-stone (details below). Although for 89% of thetime these two outlier ages are not included in themodel results (because they are down-weighted), atleast 11% of the time they still have influence onthe modelled ages and are probably responsible forthe low convergence values (<95%) generated bythe Bayesian algorithms in SU5, indicating manydifferent incompatible solutions to this section ofthe model (cf. Bronk Ramsey 1995).
The model was re-run without the outliers. Thissecond chronostratigraphic model returned statis-tically acceptable results with convergence valuesof >98% and no outliers (Table 6). The intervalbetween Phases was also calculated using the‘Interval’ command in OxCal (Bronk Ramsey2018). The results are given in Table 6 and suggestthat early in the sequence in this part of the site,campfires were made and/or landscape firesoccurred (i.e. producing wood charcoal) every c.
300 years, with more regular visitation to this partof the site fronting the Wanjina paintings after c.640 cal BP.
The results show an age sequence reflectinggood chronostratigraphic integrity. The individualcalibrated ages and Bayesian analysis each indicatethat SU6 and SU7, the deepest two layers consist-ing of massive roof-fall devoid of cultural materialsand charcoal, formed sometime before3,000–2,500 cal BP. In an adjacent area outside thedripline, Square F5 (see Figure 3) revealed charcoalat c. 125 cm below surface. It returned a calibratedage range of 3,230–3,070 cal BP (Wk-44397), con-sistent with and predating SU7. The start of SU5above the massive SU6 roof-fall layer immediatelyshows the first dated evidence of human presencein the excavation pit reported in this paper; ahearth was built on the rock surface at the junc-tion of Squares C5, C6, D5 and D6 sometimebetween 3,000–2,270 cal BP. This was followed bya period of up to 440 years duration (Interval A)when wood charcoal was not produced, and forwhich there is no evidence of human occupationin this part of the site. Evidence of human pres-ence in the form of stone artefacts associated withcharcoal is again evident near the base of SU5dated within the period 2,370 and 2,110 cal BP; itwas then that a large grindstone was positioned inplace and first used (see below). There is then asecond period without carbon ages, Interval B thatlasts 110 to 660 years, after which the bulk of SU5was laid down sometime between 2,080–1,160 calBP. Cultural materials in this main SU5 Phaseconsist of large numbers of flaked stone artefacts.This is followed by Interval C, a period of 120 to590-years duration when there are again no signsof burning. Cultural evidence then reappears at thebase of SU4, dated within 1,110–760 cal BP, afterwhich there is a fourth hiatus in radiocarbondeterminations, Interval D of 170 to 390-years’duration. The SU3 hearth layer is then dated to630–500 cal BP, immediately followed by the startof SU2 at 480–340 cal BP. SU2 then ends at450–120 cal BP with the start of sediment build upin SU1B. SU1B ends between 270 cal BP andthe present.
Each of these radiocarbon (read ‘burning’) Phasesis associated with cultural materials, and indeed insome cases at least – most demonstrably the hearthat the base of SU5, and the SU3 hearth layer – thedeposition of the charcoal itself can be shown to befrom cultural events.
Cultural materials
Cultural materials were found in all SUs exceptSU6 and SU7. They fall into six major types:
Figure 12. Northeastern wall of Squares C6, D6 and DVIafter end of excavation, prior to excavating Square D7(photo: Bruno David).
18 B. DAVID ETAL.
0cm 10 20 30 40 50 60 70 80 90 100
110
120 0cm 10 20 30 40 50 60 70 80 90 100
110
120
1
root
lets
root
arte
fact
hear
th s
tone
polle
n sa
mpl
eor
ient
ed s
edim
ent s
ampl
ero
ckch
arco
al1
40 50 60 70 80 90 100
110
120
xu1
xu2
xu3
xu4
xu5
xu6
xu7
xu8
xu9
xu10
xu11
xu12
xu13
Bxu
14B
xu15
Bxu
16B
xu15
Axu
16A
xu17
Axu
18
xu1
xu2
xu3
xu4
xu5
xu6
xu7
xu8
xu9
xu10
xu11
xu12
xu13 xu
14xu
15xu
16xu
17A
xu18
A
xu17
B
xu1
xu2
xu3
xu4
xu5
xu6
xu7
xu8
xu9
xu10
Axu
11A
xu12
xu13
xu14
xu15
Bxu
16B
xu17
Bxu
15A
xu16
Axu
17A
xu1
xu2
xu3
xu4
xu5
xu6
xu7
xu8
xu9
xu10
xu11
xu12
xu13
xu14
xu17
B
xu15
xu16
xu17
Axu
18A
xu18
Bxu
18A
xu11
B
xu18
Bxu
19A
xuxu16
B16
B6Bxu
15xu
16xu
17A7A
xuxuxu18
A18
A18
A
xuuxu17
B17
B7Bxu
14xu
15xu
16xu
17A
xuxuxu18
A18
A18
Axu
19A
quar
tz fl
ake
xu19
A
xu20
xu21
xu22
xu23
xu24
xu25
xu26
xu27
xu28
xu21
xu29
xu19
xu20
xu21
xu22
xu23
xu24
xu25
xu26
xu27
xu2
xu3
xu4
xu5
xu6
xu7
xu8
xu9
xu10
xu11
xu12
xu14
xu15
xu16
xu17
xu18
xu13 xu
21
xu22
xu19
xu20
xu1
xu23
xu24
xu25
xu26
xu27
xu19
Bxu
20xu
21
xu22
xu23
xu24
xu25
xu26
xu27
xu28
xu29
xu20
Bxu
20A
xu19
xu21
xu22
xu23
xu24
xu25
xu26
xu27
xu28
xu29
not
exca
vate
d(p
edes
tal)
not e
xcav
ated
not e
xcav
ated
not e
xcav
ated
not e
xcav
ated
not e
xcav
ated
not e
xcav
ated
C5
D5
DV
DV
ID
V
SU1A
SU1B
SU2
SU3
SU5
SU6
not
exca
vate
d(p
edes
tal)
SOUTH
WES
T
SU1A
SU1B
SU2
SU5
SU6
SU4
SU3 SU
4
3
2 1
38
8±
20
93
3±
20
10
60
±2
0
10
35
±2
01
65
8±
20
16
61
±2
01
76
9±
17
22
68
±1
6
22
50
±1
52
54
0±
15
‘pos
t-bo
mb’
378±
20
318±
20
344±
20
1
en s
ampl
eor
ient
ed s
edim
ent s
ampl
e
xu14
(pro
ject
ed)
xu15
(pro
ject
ed)
xu16
(pro
ject
ed)
xu17
(pro
ject
ed)
xu18
(pro
ject
ed)
xu1
xu2
xu3
xu4
xu5
xu6
xu7
xu8
xu9
xu10
xu11
Axu
12A
xu13
Axu
14A
xu15
Axu
16A
xu17
A
xu11
Bxu
12B
xu13
B
xu14
Bxu
15B
xu16
Bxu
17A
xu18
xu19
xu20
xu21
xu22
xu23
xu24
xu25
xu26
xu27
xu28
xu29
xu28
xu29
xu27
xu26
xu25
xu24
xu23
xu22
xu21
xu20
Axu
19
xu30
xu31 xu
32
xu33
xu34
xu35
xu37
xu26
xu27
xu1
xu2
xu2
xu3
xu3
xu4
xu4
xu5
xu5
xu6
xu6
xu7
xu7
xu8
xu8
xu9
xu9
xu10
Axu
10B
xu10
Bxu
11B
xu11
B
xu12
B
xu12
xu13
B
xu13
xu14
xu15
Axu
16A
xu17 xu
18A
xu1
xu17
B
xu1
xu2
xu3
xu4
xu5
xu6
xu7
xu8
xu9
xu10
xu11
Axu
12A
xu14
Axu
15A
xu16
Axu
17A
xu18
xu19
xu13
A
xu20
xu21
xu22
xu23
xu24
xu25
xu26
xu27
xu28
xu2
xu3
xu4
xu5
xu6
xu7
xu8
xu9
xu10
xu11
xu12
xu14
xu15
xu16
xu17
xu18
xu13
xu21
xu22
xu19
xu20
xu23
xu24
xu25
xu26
xu27
xu1
not e
xcav
ated
not e
xcav
ated
C6
DV
ID
6C
6C
5
13
11
12
10
9 8
7 6 5 4 3 2 1
4
SU5
SU6
SU7
white
pig
men
t in
D7
EAST
NORT
H
SU1A
SU1B
SU2
SU5
SU6
SU3
SU4
SU4
SU3
SU2
SU1B
SU1A
87
±2
0
22
64
±1
6
51
5±
20
46
4±
20
50
2±
20
22
59
±1
6
24
55
±1
82
23
6±
18
22
38
±1
7
22
29
±1
5
10
92
±1
7
10
08
±1
5
11
33
±1
5
45
7±
18
35
1±
16
13
0±
15
36
4±
15
35
3±
15
19
38
±1
62
20
8±
15
1099
±20
910±
20
1028
±20
959±
16
Figu
re13.Sectiondraw
ings,B
orolog
a1(artwork:Brun
oDavid
andKara
Rasm
anis).
AUSTRALIAN ARCHAEOLOGY 19
hearths, flaked stone artefacts, plant-processingbasal and top grindstones, a large grindstoneinstallation, earth pigments imported into the site,and small fallen spalls of the rock wall that containtraces of pigment and dried paint drops (Table 7,Figure 16). Stone artefacts are found throughoutthe cultural sequence except during the four radio-carbon Intervals (A–D) during which there is noevidence of charcoal deposition. In this paper, wefocus on the earth pigments and grindstones inchronostratigraphic context.
Three adjacent hearths constitute SU3. Thethickest of these, c. 15 cm thick in Squares C6-D6,was built immediately on top of the then exposedupper surface of a very large grindstone in SquareD6, so that the latter’s surface became extensivelyburnt (Figure 17). Each of these three concentratedcharcoal lenses contained burnt hearth stones ori-ented in all directions within the hearth sediments,and a single burnt flat rock up to 25 cm longplaced horizontally at the very top of the hearth,representing a distinctive act of hearth treatment.In Square C5 on the edge of the hearths, the baseof a basal grindstone lay flush with the upper sur-face of the SU3 hearth layer, and an associated
topstone lay very slightly lower next to it (Figure18). There is no stratigraphic separation betweenthe top of the hearth layer and the base of thegrindstones (charcoal particles from the hearthlayer adhere to the base of the basal grindstone,and the topstone is slightly embedded in SU3).Residue analysis of the basal grindstone’s workedsurface reveals a dense concentration of starchgrains indicating the processing of plants (resultsto be published separately).
Another, large grindstone has its base locatedwell within the base of SU5 in Square D6. Culturalmaterials including flaked stone artefacts occurthroughout SU5. Determining at exactly whichlevel in SU5, and thus when, this large 219 kgblock that measures 58 cm long, 50 cm wide by27 cm thick was put in place is not immediatelyevident, for it would have sunken slightly into theground when it was first laid down. As it was posi-tioned in place, we can expect what was then thefloor to have been scuffed to some degree. A num-ber of critical stratigraphic clues enable us to pin-point exactly when the grindstone was putin place.
The base of the grindstone became archaeologic-ally exposed in Square D6 XU28, at a depth of57.7 cm below datum. The base is flat, as is its paral-lel, crushed and ground (from the act of poundingand grinding) upper surface. The grindstone alsohas a faint reddish tinge across its upper surface,presumably from the processing of red ochre. Whilethe base of the grindstone sits in SU5, its upper sur-face lies 27 cm higher at the bottom of SU3 beingeventually fully buried with charcoal and ash fromthat major hearth (Figure 17).
The grindstone has been deliberately levelled bythe use of three small rocks, each c. 15 cm longand 10 cm high, piled one on top of the other tochock its overhanging northwest corner (Figure17). The base of the lowermost chock rock indi-cates the level at which the large grindstone waspositioned into place. That level of the base of thelowest chock rock lies in XU27, 58.4 cm belowdatum, a mere 7mm lower than the base of thelarge grindstone and therefore almost at an identi-cal level, indicating that the grindstone itself didnot penetrate much into underlying sedimentswhen it was put in place (keeping in mind that theoriginal ground surface would probably not havebeen perfectly horizontal). Also in XU27, from52.7 cm below datum – i.e. 5.7 cm above the baseof the lowest chock rock – a charcoal sample (Wk-45969) was AMS radiocarbon dated (Table 5). Itgave an age of 2,259 ± 16 BP, which has been cali-brated to 2,350-2,180 cal BP at 95.4% probability.Below this, in XU29, two charcoal samples were
Figure 14. Fire-stained rock surface representing the floor ofa hearth at the junction of Squares C5, C6, D5 and D6, baseof SU5 (photo: Bruno David).
20 B. DAVID ETAL.
Table5.
Radiocarbo
ndeterm
inations.
XUDepth
Below
Datum
(cm)
SU
Wk-
Labo
ratory
Code
d13 C&
%Mod
ern
Carbon
(F14C%)
14CAg
e(yearsBP)
Unm
odelled
CalibratedAg
eBP
(68.2%
prob
ability)
Unm
odelled
CalibratedAg
eBP
(95.4%
prob
ability)
Median
Calibrated
AgeBP
Square
C52
4.4
1B–2
interface
45959
�25.0±
0.3
98.4±0.2
130±
15270–20
270–10
110
1029.0
3–4interface
45960
�24.4±
0.2
95.6±0.2
364±
15480–330
500–320
440
1328.5
3–4interface
45961
�24.9±
0.2
95.7±0.2
353±
15470–330
490–310
400
2252.8
5–6interface
45962
�27.0±
0.2
78.6±0.2
1,938±
161,920–1,860
1,930–1,820
1,890
2358.0
5–6interface
45963
�26.2±
0.2
76.0±0.1
2,208±
152,310–2,150
2,310–2,150
2,230
Square
C68
22.7
2–3interface
45952
�23.5±
0.3
95.7±0.2
351±
16470–320
490–310
390
13b
41.7
3–4interface
45953
�24.0±
0.3
94.5±0.2
457±
18520–500
530–490
510
15a
36.0
4–5interface
45954
�25.0±
0.3
86.8±0.2
1,133±
151,070–980
1,070–970
1,020
17a
39.9
4–5interface
45955
�24.4±
0.3
88.2±0.2
1,008±
15960–920
960–910
930
1842.6
4–5interface
45958
�24.7±
0.3
87.3±0.2
1,092±
171,050–960
1,060–950
1,000
2457.8
5–6interface
45956
�25.6±
0.3
75.8±0.1
2,229±
152,310–2,180
2,330–2,150
2,220
2564.9
5–6interface
45957
�27.9±
0.3
75.7±0.2
2,238±
172,320–2,180
2,330–2,150
2,220
Square
D5
41B–2
interface
44512
n/a
149.1±
0.3
149.1±
0.4%
62
44513
�24.5±
0.3
95.4±0.2
378±
20500–330
510–320
460
102
44515
�24.8±
0.3
96.1±0.2
318±
20430–310
460–300
390
1232.6
344077
n/a
95.3±0.2
388±
20500–340
510–330
470
15b
2–3interface
44516
�27.7±
0.3
95.8±0.2
344±
20460–320
490–310
390
16a
44.4
4–5interface
44078
n/a
89.0±0.2
933±
20910–790
920–790
850
17a
46.2
4–5interface
44517
�24.8±
0.3
87.6±0.2
1,060±
20970–930
1,050–920
960
1849.2
4–5interface
44511
�25.7±
0.3
87.9±0.2
1,035±
20960–930
980–920
950
1850.6
544510
�26.5±
0.3
81.4±0.2
1,658±
201,600–1,530
1,610–1,520
1,560
1850.6
544079
n/a
81.3±0.2
1,661±
201,600–1,530
1,610–1,520
1,560
1953.3
545966
�26.1±
0.3
80.2±0.2
1,769±
171,720–1,620
1,730–1,610
1,660
2161.3
5–6interface
45967
�26.3±
0.2
75.4±0.1
2,268±
162,340–2,210
2,350–2,180
2,320
2268.2
5–6interface
45968
�23.7±
0.2
75.6±0.1
2,250±
152,330–2,180
2,340–2,160
2,230
Square
D6
28.7
1B44074
n/a
98.9±0.2
87±20
260–30
260–30
110
825.3
344075
n/a
93.8±0.2
515±
20540–520
550–510
530
11b
30.9
344508
�25.2±
0.3
94.4±0.2
464±
20520–500
530–490
510
11b
30.6
344076
n/a
93.9±0.2
502±
20540–510
550–510
530
(continued)
AUSTRALIAN ARCHAEOLOGY 21
individually dated. Wk-45970 came from 59.0 cmbelow the datum (i.e. 6mm below the lowest chockrock) and gave an age of 2,455 ± 18 BP (calibratedto 2,710–2,370 cal BP). Wk-45971 came from62.1 cm below datum (i.e. 3.7 cm below the lowestchock rock), giving an age of 2,236 ± 18 BP (cali-brated to 2,330–2,150 cal BP). While Wk-45969and Wk-45971 represent levels 5.7 cm above and3.7 cm below the base of the lowest chock rockrespectively, both give ages within the calibratedprobability range 2,350–2,150 cal BP. Wk-45970in-between is slightly older. Given the consistentage bracketing of the two younger ages, we inter-pret these results to mean that the small rockswere positioned to chock the grindstone into placesometime between 2,350-2,150 cal BP, the slightlyolder, out-of-sequence Wk-45970 age having beingscuffed up from slightly deeper sediments duringinstalment of the grindstone and its chock blocks.This is consistent with the age estimate of2,370–2,110 cal BP from the Bayesian analysis.Given that the base of the large grindstone itselfalso lies at this same general stratigraphic level, weconclude that it was put into position, oriented flatready for use, at that time.
Analysis of buried pigments
This leaves for discussion (1) the buried small frag-ments of earth pigment that were processed on-siteand that abound through the deposit, (2) the spallswith pigment and (3) the dried paint drops. Thespalls and paint drops almost certainly fell into thedeposit from the overlying Art Panel B1 rock ceil-ing. No ochre crayons – pieces of ochre with evi-dence of striations or bevelled edges – were foundin the excavated material.
Excluding the ochre-stained large grindstoneitself, in total 56 items with pigment were recoveredin situ during the excavation (Figure 19) or duringsorting of the sieved excavated sediments.
The small fragments of pigment
Fifty-two small (8–29mm long) to tiny (c. 2mmlong) angular fragments of red ochre (Figures 19and 20), two pieces of white, and two pieces of redochre with traces of white adhering to them wereexcavated. One of the pieces of white pigment wasextracted in situ from Square D7 XU17 (Figure 21).It consists of a powdery conglomerated mass with agrainy, porous texture, associated with other whitefragments that were more uniform in texture, com-pact and hard. The largest piece of red ochreweighs 2.5 g.Ta
ble5.
Continued.
XUDepth
Below
Datum
(cm)
SU
Wk-
Labo
ratory
Code
d13 C&
%Mod
ern
Carbon
(F14C%)
14CAg
e(yearsBP)
Unm
odelled
CalibratedAg
eBP
(68.2%
prob
ability)
Unm
odelled
CalibratedAg
eBP
(95.4%
prob
ability)
Median
Calibrated
AgeBP
154–5interface
44507
�26.4±
0.3
87.2±0.2
1,099±
201,050–960
1,060–960
1,000
174–5interface
44506
�24.9±
0.3
89.3±0.2
910±
20910–780
920–760
860
184–5interface
44505
n/a
88.0±0.2
1,028±
20960–920
970–920
940
2752.7
5–6interface
45969
�25.1±
0.3
75.5±0.1
2,259±
162,340–2,200
2,350–2,180
2,310
2959.0
Hearth
base
SU5
45970
�24.0±
0.2
73.7±0.2
2,455±
182,700–2,440
2,710–2,370
2,590
2962.1
5–6interface
45971
�25.6±
0.2
75.7±0.2
2,236±
182,320–2,180
2,330–2,150
2,220
Square
DV
2268.2
Hearth
base
SU5
45965
�26.8±
0.3
72.9±0.1
2,540±
152,740–2,710
2,750–2,520
2,720
Square
DVI
2268.2
5–6interface
45950
n/a
75.4±0.1
2,264±
152,340–2,210
2,350–2,180
2,320
Square
D7
1749.8–52.9
4–5interface
46419
n/a
88.7±0.2
959±
16930–800
930–790
850
Blacksamplefrom
paintedWanjinaeyeon
ceiling
ofPanelB
147814
n/a
95.9±0.2
338±
15460–310
480–310
380
All1
4 Cages
areAM
Son
sing
lepieces
ofcharcoalexcept
fortheon
eageon
charcoalrock
art.Calibratio
nsun
dertaken
usingOxCal
4.2(In
tCal13)(Reimer
etal.2
013).
22 B. DAVID ETAL.
The 56 small fragments of pigment were concen-trated in four chronostratigraphic Pigment Horizons,from top to bottom (Figure 16):
� An upper, Pigment Horizon IV, at the base ofSU1B, from 2–12 cm below ground, consistingexclusively of tiny fragments of red ochre(N¼10, total weight¼0.53 g). The Bayesian
Sequence Analysis dates this horizon after450 cal BP.
� Pigment Horizon III that corresponds with the‘SU3 hearth layer’ phase of the Bayesian SequenceAnalysis (including SU3-SU4 interface) from17.9–36.7 cm depth. It dates within 630–340 calBP. It also consists exclusively of tiny fragmentsof red ochre (N¼ 21, total weight¼ 3.98 g).
Modelled age (cal BP)
4,500 4,000 3,500 3,000 2,500 2,000 1,500 1,000
Figure 15. Bayesian age model for cultural levels at Borologa 1. The lighter shaded distributions of the individual calibrationsare the unmodelled calibrated ages. Darker outlines represent the results after Bayesian modelling (see text for discussion)(image: Fiona Petchey and Bruno David).
AUSTRALIAN ARCHAEOLOGY 23
� Pigment Horizon II (36.7–48.1 cm depth), inSU4, consisting of both red ochre fragments(N¼ 14, total weight¼1.70 g), a small piece ofred pigment with stains of white pigment on itssurface (0.15 g), and the small mass of white claypigment (0.50 g) deposited in Square D7 near theedge of the large grindstone. It corresponds withthe ‘Base of SU4’ phase (including the SU4-SU5interface) of the Bayesian Sequence Analysis,dated within 1,110–760 cal BP.
� Pigment Horizon I (47.3–67.0 cm depth) hasseven red ochre fragments (total weight¼3.50 g),one piece of red pigment with adhering trace ofwhite (0.06 g), and one white fragment (0.13 g).It corresponds with the ‘SU5’ phase of theBayesian Sequence Analysis, dated within2,080–1,160 cal BP.
These pigment fragments are interpreted as by-products of the processing of larger ochre pieces forpaint-making. As larger blocks were crushed andground, tiny pieces became scattered along the edgesof the grindstone, falling onto the surroundingground. Both red and white appear to have beenprocessed on the grindstone, as evident by the twopieces of red with adhering traces of white. Thesmall mass of white pigment is an exception: it wasplaced 15 cm from the northeastern edge of thegrindstone, when the grindstone’s flat working sur-face was still 23 cm above what was then the floor(Figure 21).
Most of the red, and white, earth pigment frag-ments were located surrounding the large grind-stone of Square D6, just outside the overhang thathas the Wanjina paintings with white background.The two excavation squares furthest from the
grindstone and from the overhanging ceiling withwhite background – Squares DV and DVI – didnot have any pieces of white pigment. All the pig-ment fragments from Pigment Horizons I–III dateto times when the large grindstone’s upper surfacewas exposed above ground and available for use.Those levels contain 46 of the 56 fragments (82%).However, the pigment from the uppermostPigment Horizon IV could not have been proc-essed from that grindstone, as its highest pointhad already been completely buried by 16 cm ofash and loamy sand by that stage. We suggest the10 tiny red ochre fragments from PigmentHorizon IV were produced either through crush-ing and grinding on a portable grindstone (sinceremoved) or from one or more of the three grind-ing patches on the flat bedrock a short distanceinside the shelter under the overhang.
The paint drops
Two samples consisting of small masses (weighing0.04 g and 0.19 g, respectively) of densely packeddisintegrating sandstone grains, with the outergrains adhered together by white pigment, wererecovered from Square C6 XU17A. They came froma part of the excavation pit located vertically underthe white-painted ceiling. We interpret these as smallpieces of disintegrating sandstone that fell from theceiling during painting, and call them ‘paint drops’because it is the dry white pigment that weakly holdsthe outer grains together (for an example of driedpaint drops excavated elsewhere in northernAustralia, see David et al. 2017). The paint dropsdate to Pigment Horizon II (within 1,110–760 calBP), the same level containing the small mass of
Table 6. Modelled results showing chronostratigraphic Phase boundary ages for the top of SU6 to SU1B.
Boundary divisionsbetween Phases
Calibrated modelled age,68.2% probability
Calibrated modelled age,95.4% probability
Interval between Phases (SU)
68.2% probability 95.4% probability
‘End Base of SU1B’ 260–0 cal BP 270–190 cal BP‘Between Base of SU1B
and SU2’350–240 cal BP 450–120 cal BP
‘Between SU2 and SU3Hearth Layer’
470–400 cal BP 480–340 cal BP
‘Start SU3 Hearth Layer’ 560–510 cal BP 630–500 cal BPDuration of Interval D 260–370 years 170–390 years‘End Base of SU4’ 910–820 cal BP 910–760 cal BP‘Start Base of SU4’ 1,060–990 cal BP 1,110–970 cal BPDuration of Interval C 370–560 years 120–590 years‘End SU5’ 1,570–1,400 cal BP 1,600–1,160 cal BP‘Start SU5’ 1,800–1,620 cal BP 2,080–1,610 cal BPDuration of Interval B 370–580 years 110–660 years‘End Base of SU5’ 2,220–2,160 cal BP 2,320–2,110 cal BP‘Start Base of SU5’ 2,320–2,200 cal BP 2,370–2,180 cal BPDuration of Interval A 120–390 years 10–440 years‘End Hearth at base of SU5’ 2,640–2,400 cal BP 2,690–2,270 cal BPRock boundary 2,820–2,540 cal BP 3,000–2,500 cal BP
Ages in bold indicate maximum start and end dates for each phase.
24 B. DAVID ETAL.
Table 7. Square A. Distribution of excavated cultural materials.
XU SU
Charcoal Burnt RockFlaked
Stone Artefacts GrindstoneWhitePigment Red Ochre
g # g # g # g # g # g
Square D51 1A 5.2 9 0.42 1Aþ1B 2.7 4 0.4 1 0.053 1Bþ2 0.3 2 0.24 1Bþ2 0.25 2 <0.16 2 0.27 2 0.1 1 <0.18 2 0.39 2þ3 0.2 1 <0.110 2þ3 0.311 2þ3 4.012 2þ3 17.1 5 1.413a 2þ3þ4 0.9 1 <0.113b 2þ3 41.0 2 <0.114a 3þ4 1.214b 3þ4 73.415a 3þ4 1.8 3 0.1 1 0.0115b 3 84.8 2 <0.116a 4 1.5 11 0.616b 3 8.3 1 0.0317a 4 3.2 41 3.7 1 0.0317b 3þ4 1.3 7 0.318 4þ5 41.8 178 45.4 5 0.2719 4þ5 0.120 5 1.321 5þ6 11.7 201 35.722 5þ623 6 <0.124 625 626 627 6Sub-total 303.1 468 88.7 9 0.39Square D61 1A 5.6 17 1.32 1B 6.8 19 1.7 4 0.093 1Bþ2 0.1 9 1.9 1 0.024 1Bþ2 0.5 1 <0.15 2 0.2 6 0.76 2 0.67 2 10.48 2þ3 1.9 1 0.029 2þ3 3.3 1 <0.110a 2þ3 4.3 3 0.310b 2þ3 73.0 3 2.2211a 2þ3 1.2 1 0.111b 3 112.3 2 0.1 2 0.3612a 2þ3þ4 2.1 1 0.412b 3þ4 0.313a 3þ4 0.4 1 <0.113b 3þ4 11.5 1 0.1 1 0.0214 3þ4 0.215 4 1.416 4 3.9 5 0.517 4þ5 12.0 6 0.7 2 0.4718 4þ5 5.3 29 2.6 1 0.1026 527 5þ6 0.228 5þ6 <0.129 5þ6 27.4 132 15.4 1� 0.0630 5þ6 4.1 35 1.731 6 <0.1 10 0.332 6 <0.1 2 0.133 634 6 1 0.235 6þ736 6þ737 7Sub-total 289.3 282 28.4 15 3.30Square DV1 1Aþ1B 5.2 5 0.82 1Bþ2 2.0 7 4.83 1Bþ2 0.2 6 0.5 1 0.244 2 0.3
(continued)
AUSTRALIAN ARCHAEOLOGY 25
Table 7. Continued.
XU SU
Charcoal Burnt RockFlaked
Stone Artefacts GrindstoneWhitePigment Red Ochre
g # g # g # g # g # g
5 2 0.36 2 0.1 1 <0.17 2 0.18 29 210 211 2 0.112 2 0.213 2 2.114 2þ3 27.2 1 0.115 2þ3þ4 20.0 4 0.216 2þ3þ4 27.1 2 <0.117a 3þ4 0.3 2 0.217b 3 86.1 1 <0.118a 3þ4 3.2 49 4.2 4 0.1018b 3 67.9 13 0.919aþb 314þ520 3þ4þ5 0.421 5 8.2 180 63.1 2 0.2122 5þ6 13.7 131 66.4 1 0.0823 5þ6 1.0 57 65.524 6 <0.1 5 0.225 626 6 1 <0.127 6 1 0.128 629 6Sub-total 265.8 466 207.3 8 0.63Square DVI1 1Aþ1B 0.2 2 0.12 1B 5.0 17 1.9 1 0.073 1Bþ2 1.0 7 0.4 1 0.024 1Bþ2 1.1 4 0.2 1 0.045 2 1.26 27 2 0.2 1 <0.18 2 0.19 2þ3 3.810a 2þ3 0.810b 2þ3 4.111a 2 <0.111b 2þ3þ4 22.312 213þ4 6.613 2þ3þ4 16.8 2 0.1 6 1.1414 2þ3þ4 31.315a 4 0.715b 2þ3 74.9 4 0.316a 4 1.7 1 <0.116b 3 44.217a 4 1.1 6 0.217b 3 71.718a 4þ518b 3 14.6 3 0.219 3þ4þ5 <0.120aþb 3þ4þ5 0.421 5 0.122 5þ623 5þ624 6 <0.1 15 3.425 6 3 0.126 6 2 0.327 6þ7 1 <0.128 729 7Sub-total 304.2 68 7.5 9 1.27Square D71 1bþ2 0.4 1 0.42 2 0.2 1 0.13 2 0.14 2 0.1 1 <0.15 2 0.16 2 0.1 1 4854.8 1 0.17 2þ3 6.8 1 0.1 1 423.28 3 8.19 3 5.4
(continued)
26 B. DAVID ETAL.
white pigment in Square D7 XU17 on the other sideof the large grindstone (described above).
The fallen spalls
Two small broken pieces of sandstone have whitepigment on one side (rather than embedded in therock), presumed to be spall flakes from the Wanjinaart panel with the white background immediatelyoverhead. Like the two paint drops described above,one (0.17 g) comes from Square C6 XU17A, theother from 4.2 cm deeper in the sandy deposit ofXU19 in the same square. They both come fromPigment Horizon II that also contains the paintdrops of Square C6 and the mass of white pigmentnear the grindstone in Square D7. Like the twopaint drops, we interpret these two small spalls ofwhite-painted sandstone as having fallen from theceiling while it was being painted (see alsoHughes 1976). They probably fell during or aroundthe time of the same painting event as the paintdrops. They consist of small spalls that were lessdisintegrated than the paint drops, so that the whitepaint remained more on the surface of thesand grains.
Pigment analyses
Geochemical analyses of excavated and potentialsource pigment samples were undertaken to charac-terise the excavated samples with a mind to eventu-ally connect the excavated pigments to the on-rockpaintings, and to sources (Table 8). We note thatthe paintings on the walls of Borologa 1 have notbeen systematically sampled for pigment analysis, soin this paper, our results are used to determine thenature and variability of pigments used, and rela-tionships with a nearby source.
Eight of the excavated red pigment, five of thewhite, and one of the red with a surface trace ofwhite pigments were analysed, including the smallmass of white excavated from Square D7 XU17(Figure 21), and a pigment source from two loca-tions on the opposite bank of the Drysdale River(Figure 22). These samples were characterisedusing X-Ray Fluorescence spectrometry (pXRF)and Field Emission Scanning Electron Microscopy(SEM) (Table 8, Figure 23). The geochemistry ofthe excavated white pigments is comparable tothat of the nearby source location on the oppositeside of the Drysdale River, while that of the redsamples is also consistent with a local geomorphicsource (see Appendix A for details of thegeochemistry).
Direct dating of a Wanjina painting
An AMS radiocarbon age was also obtained on asmall, exfoliating fragment of the left, black eye ofa Wanjina painting on Art Panel B1 adjacent tothe excavation squares. Such sampling of Wanjinamotifs is rarely allowed because of their ongoingsignificance to local Traditional Owners. In thiscase, permission was granted due to the collabora-tive nature of the research and mutual interest inthe results. The grey pigment of the eye showsclear striations indicating application by brush.Denser, black fragments across the eye and its out-line suggest that the eye was originally paintedblack, but with time this pigment has mostly fallenoff, leaving the underlying layer of white pigmentstained grey. The eye was encircled by a red linethat partially overlies the grey, but there are noclear instances of its overlying any of the remain-ing denser black regions. The ‘halo’ of the Wanjinahas been repainted and slightly altered in its
Table 7. Continued.
XU SU
Charcoal Burnt RockFlaked
Stone Artefacts GrindstoneWhitePigment Red Ochre
g # g # g # g # g # g
10 3þ4 1.511 3þ4 0.212 4 0.113 4 0.3 2 0.1 1 0.0514 4 0.7 3 0.215 4þ5 0.8 6 0.216 4þ5 1.4 15 7.117 4þ5 5.2 29 3.5 1 0.50 3 0.3618 5 2.9 25 2.4 1� 0.1519 5 1.1 14 1.020 5þ6 0.1 3 0.321 5þ6 <0.122 6Sub-total 35.7 1 4854.8 102 15.6 1 423.2 1 0.50 5 0.56Total 1268.5 1 4854.8 1813 397.0 1 423.2 1 0.50 46 6.15
XUs highlighted in grey list the saved in situ cultural materials only: the sieved materials were lost in the helicopter-drop and could not, therefore,be included (Squares C5 and C6 are omitted as most materials were lost in that accident).�Red pigment with trace of white.
AUSTRALIAN ARCHAEOLOGY 27
Table8.
Semiq
uantitativeelem
entabun
dances
measuredby
pXRF
(Net
Peak
Areasof
Klines).
Spectra#
SampleNam
eCo
lour
Mg
AlSi
PS
ClK
CaTi
Mn
Fe
Square
C5186
XU14
Red
Not
Calc
8,925
25,926
198
359
2,576
109,644
1,474
33,047
1,665
945,493
178
XU19
(Sam
ple1SurfaceA)
Red
149
10,496
33,027
532
358
2,654
129,066
1,917
53,718
2,252
3,269
179
XU19
(Sam
ple1SurfaceB)
Red
149
9,730
27,202
606
375
2,676
141,215
1,996
55,959
2,124
4,058
168
XU22
(Sam
ple1)
White
167
7,066
20,549
127
347
2,383
86,251
1,387
25,378
590
173,647
166
XU22
(Sam
ple3SurfaceA)
White
136
6,946
22,089
170
326
2,404
87,820
1,699
26,136
701
158,079
167
XU22
(Sam
ple3SurfaceB)
White
877,118
22,312
143
330
2,576
79,947
1,479
25,090
479
143,040
170
XU22
(Sam
ple4)
White
145
6,648
20,350
157
355
2,522
80,881
1,605
24,760
595
141,504
171
XU22
(Sam
ple5)
White
836,852
21,304
148
330
2,434
77,781
1,410
24,411
499
104,507
176
XU23
(Sam
ple1SurfaceA)
Red
9211,822
39,382
306
366
2,684
142,899
1,896
53,304
2,175
1,691
177
XU23
(Sam
ple1SurfaceB)
Red
182
11,640
37,527
265
353
2,886
157,962
1,881
51,481
1,918
1,437
181
XU23
(Sam
ple2SurfaceA)
Red
123
9,609
34,929
341
382
2,747
120,893
1,947
50,543
2,477
1,261
182
XU23
(Sam
ple2SurfaceB)
Red
109
9,356
30,316
308
390
2,813
136,749
1,842
48,482
2,147
1,253
180
XU23
Red
Not
Calc
6,070
18,413
137
359
2,615
82,342
1,377
33,803
648
1,097,480
Square
C6192
XU17A
White
onsand
ston
espall
587,131
64,725
2,387
314
2,477
55,672
1,457
28,936
182
142,989
204
XU19
White
onsand
ston
espall
23824
3,180
1851
275
8,415
153
2,817
3012,233
196
XU19
(Sam
ple4SurfaceA)
Red
137
10,510
57,980
209
315
2,393
111,209
1,804
37,410
215
578
197
XU19
(Sam
ple4SurfaceB)
Red
147
12,426
60,809
160
392
2,533
134,633
1,598
47,928
225
755
Square
D5
228
XU15a
Red
Not
Calc
4,942
16,242
258
431
2,456
17,109
2,687
41,217
1,035
2,081,313
229
XU16b
Red
Not
Calc
5,282
27,999
233
359
2,464
49,240
1,359
41,462
510
507,547
230
XU18
Red
Not
Calc
8,030
30,401
675
395
2,623
99,446
2,456
46,393
1,707
3,460,542
Square
D6
216
XU2
Red
Not
Calc
8,599
48,834
322
339
2,609
98,824
2,393
45,448
2,013
1,281,683
213
XU3
Red
Not
Calc
3,538
17,196
245
429
2,478
38,805
1,647
28,286
895
1,075,620
211
XU8
Red
Not
Calc
7,670
22,481
213
397
2,431
102,358
1,660
30,327
1,599
883,730
210
XU11b
Red
Not
Calc
12,164
33,009
260
377
2,681
175,878
1,813
40,613
3,136
1,661,393
215
XU11b
Red
Not
Calc
8,203
26,819
259
437
2,510
108,009
1,722
33,992
2,086
1,096,686
212
XU13b
Red
Not
Calc
8,434
26,835
168
485
2,504
86,654
1,727
38,343
720
1,445,441
174
XU29
(Sam
ple1redsurface)
Redwith
white
13,084
6,156
273
506
2,425
3,612
1,112
24,654
482
205
173
XU29
(Sam
ple1white
surface)
Redwith
white
2110,065
19,333
96304
1,847
4,279
807
16,109
247
Square
D7
185
XU13
Red
Not
Calc
4,333
19,456
181
350
2,439
44,659
1,351
26,774
555
448,972
187
XU17
Red
Not
Calc
7,704
28,946
238
341
2,522
78,793
2,031
47,125
8,654
3,764,370
164
XU17
(Sam
ple1)
White
665,837
14,579
6,827
406
2,449
44,871
1,936
45,480
3,490
109,022
165
XU17
(Sam
ple2)
White
8911,012
35,763
12,006
291
2,647
73,275
2,489
59,087
3,113
148,812
183
XU18
(Sam
ple1redsurface)
Redwith
white
257
12,138
21,830
13,674
1,207
2,870
93,155
34,890
128,315
36,222
15,689
184
XU18
(Sam
ple1white
surface)
Redwith
white
162
13,442
34,155
10,486
997
2,602
95,677
33,591
120,161
37,576
16,661
(continued)
28 B. DAVID ETAL.
position. The body outline has been similarlyaltered from an earlier version (Figure 24(B)).
The black eye sample consists of charcoal whosestructure could be made out under magnification. Itgave a radiocarbon determination of 338 ± 15 BP(Wk-47814), calibrated to 460–310 cal BP(median¼380 cal BP) at 68.2% probability (Table 5).Superimposed over this Wanjina image is a simplesickle-shaped non-figurative element in a very darkred-brown colour, so the radiocarbon determinationfor the black eye reveals both the age of a late paint-ing event for the eye of the Wanjina, and a max-imum age for the superimposing (i.e. later)repainting of its red halo and the re-alignment ofthe body. It also represents a minimum age for theunderlying white pigment.
Phase SU2 of the excavation sequence dates tobetween 470 and 240 cal BP. The Phase SU3 hearthlayer dates to between 560 and 400 cal BP. The agerange covered by Wk-47814 falls between 460 and310 cal BP, so could conceivably fall within eitherexcavation phase given the limited age controlobtained from a single date obtained directly fromthe rock painting of a Wanjina, but provides thegreatest statistical match with Phase SU2. As thedated black charcoal from the eye superimposes(and is, therefore, younger than) the most recent ofthe two white paint layers on the ceiling, the radio-carbon result for the black of the eye is consistentwith the evidence for white painting activityrevealed by the buried fragments of white pigment,including the two dried white paint drops and thetwo small rock spalls with white pigment, almost allof which come from Pigment Horizon II (datedwithin 1,110–760 cal BP). Some small white pigmentsamples from the earlier Pigment Horizon I (datedwithin 2,080–1,160 cal BP) indicate at least twophases of white painting near Art Panel B1, consist-ent with the two layers of white observed on theceiling. These results push back in time by a fewhundred years what we previously knew of the ageof white backgrounds associated with Wanjinapaintings, which first appear at Borologa 1 by atleast 1,160 cal BP and possibly as early as2,080 cal BP.
Conclusion
Despite its considerable interest, rock art oftenremains poorly integrated into standard archaeo-logical narratives. This is often because the age ofthe art is poorly understood, and the Kimberley isno exception (but see, for example, Green et al.2017; Ross et al. 2016). Understanding the chaıneop�eratoire or operational sequence involved in theproduction of paint, from procurement toTa
ble8.
Continued.
Spectra#
SampleNam
eCo
lour
Mg
AlSi
PS
ClK
CaTi
Mn
Fe
Square
DV
218
XU18a
Red
Not
Calc
6,596
25,800
163
398
2,634
64,005
1,669
37,698
464
2,290,257
221
XU18a
Red
Not
Calc
4,684
18,240
218
427
2,482
37,279
1,661
29,256
368
1,631,103
188
XU21
Red
Not
Calc
9,476
26,708
384
431
2,665
85,197
3,115
45,962
2,566
3,558,326
175
XU22
Red
Not
Calc
8,287
50,088
294
420
2,459
57,506
1,501
34,765
744
1,000,690
Square
DVI
232
XU2
Red
Not
Calc
7,731
35,960
434
397
2,869
100,062
2,623
49,918
2,481
3,599,733
234
XU3
Red
Not
Calc
4,681
19,840
190
475
2,440
55,703
1,578
26,807
463
846,232
217
XU4
Red
Not
Calc
3,905
18,514
306
395
2,417
43,671
1,441
28,084
820
1,029,058
233
XU13
Red
Not
Calc
9,907
33,111
793
343
2,510
1,33,244
2,163
49,898
2,727
1,998,142
DrysdaleRiverSource
236
Locatio
n1(Sam
ple1)
White
137
11,194
42,355
169
328
2,468
130,511
2,201
50,118
2,861
363,414
235
Locatio
n1(Sam
ple2)
White
223
13,878
50,883
320
391
2,605
146,051
2,305
53,440
2,091
248,612
238
Locatio
n1(Sam
ple3)
White
160
10,613
43,711
246
398
2,553
125,160
2,215
43,085
1,701
243,630
237
Locatio
n1(Sam
ple4)
White
828,582
37,910
224
390
2,480
95,933
2,236
44,446
1,373
178,543
239
Locatio
n1(Sam
ple5)
White
123
9,961
44,201
147
295
2,436
108,932
2,061
48,969
1,878
232,199
Bold
values
indicate
elevated
abun
dances.Italic
values
givenforincompleteassays
where
yieldstatisticsweretoolow
forfull90-secon
dcoun
ts,b
utqu
alitativechem
ical
profileswereob
tained.Shadedcells
indicate
thewhite
pig-
mentfoun
dnear
thegrindstone
(paint
drop
sweretoosm
allfor
analysis).
AUSTRALIAN ARCHAEOLOGY 29
3.2
1.4
2.2
White pain drops or white pigment on spallsRed & white pigmentWhite pigmentRed pigment
C5 C6 C6 D5 D6 D7 DV DVI
Weight (g)
Number of fragments
0 0.15 0 0 0 0 0 0 00.40 0.50 0.50 0.500.30 0.30 0.30
0 5 0 0 0 0 0 0 01 3 4 45 4 6
Figure 16. Distribution of pigment by excavation square (image: Jerome Mialanes and Bruno David).
Figure 17. Large grindstone in Square D6, showing its in situ horizontal positioning by the two upper of three chock rocks atits northwestern end (photos: Bruno David).
30 B. DAVID ETAL.
processing to application, is also of considerablevalue to archaeology. A common theme is that thesuccessful investigation of each of these – dating,sourcing, production sequence – requires theretrieval of fine-grained information on the pig-ments at stake. For wall paintings at Borologa 1Art Panel B1, for example, the presence of super-imposed Wanjina with the same but slightly offsetorientation as the underlying ones suggests thatthe aim may not have been only to revitalise theold image by repainting or hiding it, but also tonewly highlight and perform its revitalisation. Forburied samples, the retrieval of fine-grained infor-mation means the employment of finer excavationmethods than is often applied, with a careful eyeto the strategic positioning of excavation squares,as well as to the retrieval of tiny fragments frompigment processing and potentially of dried paintdrops from the application of paint on rock wallsor ceilings (see also McNiven et al. 2009). Suchdried paint drops, however, will only be seen insitu with finer-grained excavation methods thancurrently used by most researchers. In short, exca-vation methods need to be tuned to the recoveryof tiny samples produced during processing andpaint application, such as the crushing and grind-ing of ochre, and the residues of painting events(paint drops and roof spalls).
At Borologa 1, the structurally recognisablehearth at the base of SU5 dates between3,000–2,500 cal BP and 2,690–2,270 cal BP, andthe one that makes up SU3 dates within630–500 cal BP. The large grindstone was installedbetween 2,370–2,180 cal BP and 2,320–2,110 calBP, although its upper (grinding) surfaceremained exposed and usable above the graduallyaccumulating sediments surrounding it until
630–500 cal BP, when it finally became covered bythe SU3 hearth in Square D6. Dozens of tiny pig-ment fragments were recovered around the large,deliberately installed and levelled grindstone.Here the pigment fragments are by-products ofthe crushing and grinding of red ochre and (in alllikelihood locally extracted) white clay in themaking of paint. However, their value is morethan the accidental debris of manufacture. Rather,they are by-products that enable us to ventureinto an archaeology of the painting event itself,providing insights into dating the creation of art-works and understanding their periodic retouching,the evidence of their enduring cosmological sig-nificance. Dried paint drops under Art Panel B1’spainted ceiling also testify to the event, thechronological pattern strengthened by the contem-poraneity of the drops and white fragments asindicated by their exact chronostratigraphic corres-pondence across excavation squares on either sideof a large grindstone.
It is notable that, while the large, installedgrindstone that furnished the part of the site adja-cent to the painted ceiling was a focal point forthe production of pigment during the early phasesof paint production, during Pigment Horizon IV,it would already have been buried and could nothave been used. Rather, the upper horizon of tinyfragments of red ochre signals either that a port-able grindstone was once used and then removed,leaving behind on the ground the by-products ofthis latest episode of paint production, or that tinyfragments of red ochre became scattered during
Figure 18. Plant processing basal grindstone beforelifting (its base just exposed, flush with the top of the SU3hearth layer) and topstone (partly buried, middle of left-hand side) from Square C5, excavation in progress (photo:Bruno David).
Figure 19. Tiny fragments of red ochre (identified in whitecircles) exposed by the sweep of the trowel in Square D5XU17, excavation in progress. The finer scale is in millimetreincrements (photo: Bruno David).
AUSTRALIAN ARCHAEOLOGY 31
use of flat bedrock surfaces for the processing ofochre even closer to the painted ceiling. Eitherway, buried grindstones are not necessary for the
recovery of paint by-products as evidence of paint-ing events.
Last but not least, while the white background ofthe Wanjina panel at Borologa 1 began to bepainted sometime between 2,080–1,160 cal BP, wecan also now determine that it is during the pastmillennium that the most intensive painting andrepainting of Wanjina motifs took place here. Givena periodic but unbroken chain of Wanjina artworkson rock over this time frame both at Borologa 1and more broadly in the Kimberley, we can reason-ably extend an approximation of key elements ofthe ethnographically known cosmological significa-tion of Wanjina motifs back for many hundreds ofyears into history, with earlier, antecedent versionsalso apparent beyond c. 1,000 years ago. Today andin recent times, Wanjina spirit-beings express veryparticular local Aboriginal world views concerningthe creative powers of ancestral beings and relating
E HF GB C DA
J K LG H IF
A B C D E
UPPER PIGMENT PHASE: 450-120 cal BP to Present (= within the past ca. 500 years)
MIDDLE PIGMENT PHASE: 630-500 to 480-340 cal BP (= between ca. 700-550 and 550-400 years ago)
LOWER PIGMENT PHASE: 2,080-1,610 to 1,600-1,160 cal BP (= between ca. 2,150-1,700 and 1,650-1,250 years ago)
MLKJIH
A B C D E F G
0 10
mm
Figure 20. Examples of excavated small fragments of red ochre from Borologa 1 (photos: Steve Morton).
Figure 21. White pigment in XU17 of Square D7, immedi-ately after exposure by sweep of the trowel (photo:Bruno David).
32 B. DAVID ETAL.
people to ancestors, social (clan) affiliations, Law (asgiven by the ancestors at the beginning of time) andCountry. We suggest that being able to track funda-mental elements of Wanjina symbolism in the rockart back for 1000 years or more, signals long-lastingcontinuity in their associated cosmological associa-tions. Yet such continuity of symbolic practice alsoneeds to be considered in a longer sequence ofdynamic change in Kimberley artistic practice, inter-regional relations, intensities of landscape engage-ment and wide-ranging geographical influences.These patterns and trends will be revealed better by
a clearer understanding of the antiquity of each art-work’s individual conventions and overall style. Aswe have demonstrated at Borologa 1, this will argu-ably require archaeological field methods aimed atconnecting the buried archaeological evidence withthe art on the exposed rock walls.
Acknowledgements
We thank Augustine Unghango and BalanggarraTraditional Owners for permission to work on theirlands and partnership in doing so. For financial and
Figure 22. White pigment source on opposite side of Drysdale River to Borologa 1. Top: White pigment source under low-hanging rocks in foreground, with Borologa 1 on opposite side of river. Bottom: Close-up of white pigment source under lowrock overhang in river bed. The white pigment layer was exposed by hand by sweeping aside the unconsolidated overlyinglayer of alluvial sand (photos: Bruno David).
AUSTRALIAN ARCHAEOLOGY 33
logistical support, we thank the Australian ResearchCouncil’s ‘Kimberley Visions: Rock Art Style Provincesof North Australia’ project (LP150100490), the MonashIndigenous Studies Centre at Monash University, theUniversity of Western Australia, Maria and CeciliaMyers, Susan Bradley, Kimberley Foundation Australia,
Dunkeld Pastoral Co. Pty. Ltd., the staff at Theda andDoongan Stations. We thank the excavation and rockart recording teams: Isaac Barney, Frank Boulden,Leigh Douglas, Adrian French, Paul Hartley, BrigidHill, Lucas Karadada, Madeleine Kelly, Lorraine Lee,Robin Maher, William Maraltidj, Michael Morlumbin,
(A) (B)
(C) (D)
(E) (F)
Figure 23. SEM backscattered micrograph of pigment samples. (A, B) From Location 1 in the Drysdale River Source (A: sub-sample 2; B: subsample 5). Typical Ti enrichment (lighter grains in the centre of A) surrounded by a typical clay (Si Al) matrix.The larger, slightly lighter matrix grains in the matrix contain more Si (�18%) than the darker smaller grained sections(�16%). Very small, bright grains are the Ti, Fe, Cu and Zr grains. (C, D) From Location 2 in the Drysdale River Source. TypicalTi enrichment (lighter grains in the top right of C and centre on D) with typical clay (Si Al) matrix, the larger lighter matrixgrains contain more Si (�18%) than the darker smaller grained sections (�16%). Very small bright grains are the Ti, Fe, Cuand Zr grains. (E, F) From Square C5 XU22 (white) pigment fragment numbers 1 (E) and 3 (F). The very small bright grains arerich in Ti, Fe, Cu and Zr (image: Jillian Huntley).
34 B. DAVID ETAL.
Ken Mulvaney, Nick Sundblom, Patrick Tataya, GarethUnghango, Jeremy Unghango, Scott Unghango, IanWaina, Rowan Waina and Uriah Waina. Thanks toLaura Jacobs for her assistance with pXRF andSEM analyses during her recently completed internshipat Griffith University; Richard Cosgrove, Judith Fieldand Adelle Coster for starch analysis on the grind-stones; and Liam Brady for comments on an earlierdraft of the manuscript. Australian Research CouncilLaureate Grant FL160100123 supported JH’s participa-tion in this research; we thank Professor Paul Tacon.Scanning Electron Microscopy was undertaken at theCentral Analytical Research Facility, QueenslandUniversity of Technology, and the particle size analysesand pH measurements were undertaken at MonashUniversity, mostly by Ursula Pietrzak and Chris Urwin.Thanks to three anonymous referees for use-ful comments.
Disclosure statement
No potential conflict of interest was reported bythe authors.
Funding
This study was funded by the Australian ResearchCouncil’s ‘Kimberley Visions: Rock Art Style Provinces ofNorth Australia’ project (LP150100490); KimberleyFoundation Australia; Dunkeld Pastoral Co. Pty. Ltd.
ORCID
Bruno David http://orcid.org/0000-0002-8567-6135Peter Veth http://orcid.org/0000-0002-1717-6390Jerome Mialanes http://orcid.org/0000-0002-6790-7542
References
Akerman, K. 2016 Wanjina: notes on some iconicAncestral Beings of the northern Kimberley. Carlisle:Hesperian Press.
Allen, M.S. and R. Wallace 2007 New evidence from theEast Polynesian gateway: substantive and methodo-logical results from Aitutaki, southern Cook Islands.Radiocarbon 49(3):1–17.
Aubert, M. 2012 A review of rock art dating in theKimberley, Western Australia. Journal of ArchaeologicalScience 39(3):573–577.
Balanggarra Aboriginal Corporation and Kimberley LandCouncil 2011 Balanggarra Healthy Country Plan2012–2022 48. Retrieved 1 August 2018 from <http://www.klc.org.au/docs/default-source/Ranger-Fact-Sheets/balanggarra-healthy-country-plan-2012-2022?sfvrsn¼4>
Blundell, V., K. Doohan, D. Vachon, M. Allbrook, M.A. Jebband J. Bornman 2017. Barddabardda Wodjenangorddee:we’re telling all of you. Fremantle: Fremantle Press.
Bronk Ramsey, C. 1995 Radiocarbon calibration and ana-lysis of stratigraphy: the OxCal program. Radiocarbon37(02):425–430.
Bronk Ramsey, C. 2009 Bayesian analysis of radiocarbondates. Radiocarbon 51(1):337–360.
Bronk Ramsey, C. 2018 OxCal Program v4.3. Radiocarbonaccelerator unit. Oxford: University of Oxford.
Bronk Ramsey, C., M. Dee, S. Lee, T. Nakagawa and R.Staff 2010 Developments in the calibration and model-ling of radiocarbon dates. Radiocarbon 52(03):953–961.
Brumm, A., M.C. Langley, M.W. Moore, B. Hakim, M.Ramli, I. Sumantri, B. Burhan, A.M. Saiful, L. Siagian,R.S. Suryatman, A. A. Jusdi, A.P. Mubarak, H. Hasliana,A.A. Oktaviana, S. Adhityatama, G.D. van den Bergh, M.Aubert, J. Zhao, J. Huntley, B. Li, R.G. Roberts, E.W.Saptomo, Y. Perston and R. Gr€un 2017 Early humansymbolic behavior in the Late Pleistocene of Wallacea.Proceedings of the National Academy of Sciences 114(16):4105–4110.
Capell, A. 1972 Cave painting myths: northern Kimberley.Oceania Linguistic Monographs 18. Sydney: Universityof Sydney.
Chalmin, E. and J. Huntley 2017 Characterizing rock artpigments. In B. David and I. J. McNiven (eds), TheOxford handbook of the archaeology and anthropologyof rock art. Oxford: Oxford University Press.
Figure 24. Sample of charcoal pigment collected for AMSradiocarbon dating from a painted black Wanjina eye,Borologa 1. A: Close-up of radiocarbon-dated area after sam-pling. B: Dated Wanjina in relation to neighbouring paintings(photos: Robert Gunn).
AUSTRALIAN ARCHAEOLOGY 35
Chippindale, C. and P. S. C. Tacon (eds) 1998 The archae-ology of rock-art. Cambridge: Cambridge University Press
Clarke, J. 1976 Two Aboriginal rock art pigments fromWestern Australia: their properties, use and durability.Studies in Conservation 21:134–142.
Crawford, I.M. 1968 The art of the Wandjina: Aboriginalcave paintings in Kimberley, Western Australia.Melbourne: Oxford University Press.
David, B., J.-J. Delannoy, R. Gunn, E. Chalmin, G.Castets, F. Petchey, K. Aplin, M. O’Farrell, I. Moffat, J.Mialanes and J.-M. Geneste 2017 Dating painted PanelE1 at Nawarla Gabarnmang, central-western ArnhemLand plateau. In B. David, P. S. C. Tacon, J.-J.Delannoy and J.-M. Geneste (eds), The archaeology ofrock art in western Arnhem Land, Australia, pp.245–302. Terra Australis 47. Canberra: ANU Press.
Donaldson, M. 2012 Kimberley rock art 2: northKimberley. Mt Lawley: Wildrocks Publications.
Elkin, A.P. 1930 Rock paintings in north-west Australia.Oceania 1(3):257–279.
Ford, B., I. MacLeod and P. Haydock 1994 Rock art pig-ments from the Kimberley region of Western Australia:identification of the minerals and conversion mecha-nisms. Studies in Conservation 39(1):57–69.
Green, H., A. Gleadow, D. Finch, J. Hergt and S. Ouzman2017 Mineral deposition systems at rock art sites,Kimberley, northern Australia – field observations.Journal of Archaeological Science: Reports 14:340–352.
Hughes, P.J. 1976 Inferred rates of weathering in sandstoneshelters in southern NSW: some implications for thepreservation of rock art. In C. Pearsons and G. L. Pretty(eds), Proceedings of the 1st national seminar on the con-servation of cultural material. Perth, 1973. Sydney:Institute for the Conservation of Cultural Materials.
Huntley, J. 2015 Looking up and looking down: pigmentchemistry as a chronological marker in the SydneyBasin rock art assemblage, Australia. Rock Art Research32(2):131–145.
Huntley, J., M. Aubert, J. Ross, H.E.A. Brand and M.J.Morwood 2015 One colour (at least) two minerals: astudy of mulberry rock art pigment and a mulberrypigment ‘quarry’ from the Kimberley, northernAustralia. Archaeometry 57(1):77–99.
Huntley, J., H.E.A. Brand, M. Aubert and M.J. Morwood2014 The first Australian Synchrotron powder diffrac-tion analysis of pigment from a Wandjina motif in theKimberley, Western Australia. Australian Archaeology78(1):33–38.
Johnson, J. 2014 Accurate measurements of low Z ele-ments in sediments and archaeological ceramics usingportable X-ray fluorescence (PXRF). Journal ofArchaeological Method and Theory 21(3):563–588.
Lommel, A. 1952 The Unambal: a tribe in northwestAustralia. Carnarvon Gorge: Takarakka Nowan KasPublications. (1997 translation by I. Campbell).
McNiven, I.J., L.M. Brady and A.J. Barham 2009 KabadalKula and the antiquity of Torres Strait rock art.Australian Archaeology 69:29–40.
Morwood, M.J., G.L. Walsh and A.L. Watchman 2010AMS radiocarbon ages for beeswax and charcoal pig-ments in north Kimberley rock art. Rock Art Research27(1):3–8.
Mowaljarlai, D., and J. Malnic 1993 Yorro Yorro: spirit ofthe Kimberley. Broome: Magabala Books.
O’Connor, S., A. Barham and D. Woolagoodja 2008Painting and repainting in the West Kimberley.Australian Aboriginal Studies 1:22–38.
Ouzman, S., P. Veth, C. Myers, P. Heaney and K.Kenneally 2017 Plants before animals? Aboriginal rockart as evidence of ecoscaping in Australia’s Kimberley.In B. David and I. J. McNiven (eds), The Oxford hand-book of the archaeology and anthropology of rock art.Oxford: Oxford University Press. doi:10.1093/oxfordhb/9780190607357.013.31
Reimer, P.J., E. Bard, A. Bayliss, J.W. Beck, P.G.Blackwell, C. Bronk Ramsey, C.E. Buck, H. Cheng, R.L.Edwards, M. Friedrich, P.M. Grootes, T.P. Guilderson,H. Haflidason, I. Hajdas, C. Hatt�e, T.J. Heaton, D.L.Hoffmann, A.G. Hogg, K.A. Hughen, K.F. Kaiser, B.Kromer, S.W. Manning, M. Niu, R.W. Reimer, D.A.Richards, E.M. Scott, J.R. Southon, R.A. Staff, C.S.M.Turney and J. van der Plicht 2013 IntCal13 andMarine13 radiocarbon age calibration curves, 0-50,000years cal BP. Radiocarbon 55(04):1869–1887.
Roberts, R.,. G. Walsh, A. Murray, J. Olley, R. Jones, M.Morwood, C. Tuniz, E. Lawson, M. Macphail, D.Bowdery and I. Naumann 1997 Luminescence datingof rock art and past environments using mud-waspnests in northern Australia. Nature 387(6634):696–699.
Ross, J., K. Westaway, M. Travers, M.J. Morwood and J.Hayward 2016 Into the past: a step towards a robustKimberley rock art chronology. PLoS ONE 11(8):1–33.
Schiffer, M.B. 1987 Formation processes of the archaeo-logical record. Albuquerque: University of New MexicoPress.
Schulz, A.S. 1956 North-west Australian rock paintings.Memoirs of the National Museum of Victoria 20:7–43.
Thomas, A.M. 1998 Spirit of the serpent: an ethnographic andscientific analysis of the white pigments used in Wandjinarock art, Kimberleys, Western Australia. UnpublishedBA(Hons) thesis, Department of Archaeology andPaleoanthropology, University of New England, Armidale.
Veth, P., C. Myers, P. Heaney, and S. Ouzman 2018Plants before farming: the deep history of plant-useand representation in the rock art of Australia’sKimberley region. Quaternary International 489:26–45.
Walsh, G.L. 2000 Bradshaw art of the Kimberley.Toowong: Takarakka Nowan Kas Publications.
Ward, I., A.L. Watchman, N. Cole and M.J. Morwood2001 Identification of minerals in pigments fromAboriginal art in the Laura and Kimberley regions,Australia. Rock Art Research 18(1):15–23.
Watchman, A.L. 1987 Preliminary determinations of theage and composition on mineral salts on rock art sur-faces in the Kakadu National Park. In W. R. Ambroseand J. M. J. Mummery (eds), Archaeometry: furtherAustralasian studies, pp. 36–42. Canberra: AustralianNational University.
Watchman, A.L. 1997 Kimberley and Hann River paintand accretion compositions. Unpublished report pre-pared for the Australian Institute of Aboriginal andTorres Strait Islander Studies, Canberra.
Watchman, A.L. 2000 Micro-excavation and laser extrac-tion methods for dating carbon in silica skins and oxal-ate crusts. In G. Ward and C. Tuniz (eds), Advances indating Australian rock markings: papers from the firstAustralian rock picture dating workshop, pp. 35–39.Occasional AURA Publication 10. Melbourne:Australian Rock Art Research Association.
Watchman, A.L., B. David, I.J. McNiven and J. Flood2000 Micro-archaeology of engraved and painted rocksurface crusts at Yiwarlarlay (the Lightning Brotherssite), Northern Territory, Australia. Journal ofArchaeological Science 27(4):315–325.
36 B. DAVID ETAL.
Welch, D. 1993 The early rock art of the Kimberley,Australia: developing a chronology. In J. Steinberg, A.Watchman, P. Faulstich and P. S. C. Tacon (eds), Timeand space: dating and spatial considerations in rock-artresearch, pp. 13–21. Occasional AURA Publication 8.Melbourne: Australian Rock Art Research Association.
Welch, D. 2015 Aboriginal paintings of Drysdale RiverNational Park, Kimberley, Western Australia.Coolalinga: David M. Welch.
Appendix A: Geochemical Analyses ofExcavated and Source Pigments
A powerful application of portable X-Ray Fluorescencespectrometry (pXRF) in rock art research has been toidentify qualitatively the composition of mineral pigmentsbased on their major chemical constituents. The proced-ure involves positing at least two chemically distinctivealternatives for mineral pigments of the same colour.Following this premise, several studies have distinguishedbetween chemically discrete pigment sources (Brummet al. 2017; Chalmin and Huntley 2017; Huntley 2015).
pXRF analyses were conducted on excavated pigmentsfrom Borologa and from the nearby source (see text fordetails of individual samples) using a Bruker Tracer 5ipXRF equipped with an 8lm Beryllium, rhodium tubeand 20mm2 silicon drift detector (resolution of <140 eVat 250,000 counts per second on the Mn Ka line). Assayswere collected over 90 live seconds using an 8mm colli-mator at 15 kV and 39 lA under vacuum (<20 psi) toremove air from the beam path, with the addition of a25 lm Ti filter (as measurement of Cl and S was desir-able). Relative element abundances were calculated usingArtax (V8) software following the manual interrogationof each spectrum (Table 8). Certified reference materialswere included in analytic program to monitor instrumentstability (Johnson 2014). Almost all the excavated speci-mens analysed were smaller than the pXRF instrumentaperture (<5mm diameter), which has resulted in lowerfluorescent yields for specimens and hence lower elementabundances. Field emission SEM was undertaken on aJoel 7100 JSM. Energy Dispersive Spectrometry (EDS) ofrelative element abundances measured in spot, area andelement mapping modes using an Oxford X-Max 80probe and Aztec software. Two subsamples from differentlocations of a white pigment source located across theriver from Borologa 1, and one example of white pigmentfrom the excavation (Square C5 XU22) were embedded inepoxy resin and exposed in cross-section prior to goldcoating for investigation.
Across the Kimberley where Wanjina paintings havepreviously been characterised, white paints have beenfound to be predominantly clay (kaolinite, sometimesincluding micas such as muscovite) or huntite minerals(Chalmin and Huntley 2017; Clarke 1976; Huntley et al.2014). Chemically, clays are principally silicon and alu-minium, around 20% by weight each in kaolinite(Al2Si2O5(OH)4) and muscovite (KAl2(AlSi3O10)(FOH)2).In contrast, huntite is composed of around 11% calciumand 20% magnesium by weight (CaMg3(CO3)4).Differentiating these minerals by pXRF is thereforestraightforward – clays display large Al and Si peaks(including K in micas), whereas huntite has large Capeaks and Mg.
pXRF analysis of white pigment specimens recoveredfrom the Borologa 1 excavations returned spectra
dominated by aluminium and silicon, indicating clay con-stituents for these pigments (Table 8). Calcium and ironare consistently present in low to moderate abundance.Titanium was persistently measured (above the peak gen-erated by the instrument filter). pXRF readings indicatethe absence of weathering products such as geologicalsalts (sulphur and chlorine each having persistently lowreadings). Elevated phosphorous abundances were foundin the small mass of white pigment from Square D7XU17 and the red fragment with white paint residuefrom Square D7 XU18 (both from Pigment Horizon II), atantalising possibility that organic binding media mayhave been used in composite paint recipes, although thisrequires further investigation.
SEM analyses of two white pigments from Square C5XU22 (Pigment Horizon I) and four samples from twolocations in the white pigment source showed they are allpredominantly clay constituents (high relative abundancesof Al and Si, �18% and 13% by weight respectively). EDSanalyses also revealed a consistently low presence of mag-nesium (0.2 to 0.5% by weight, with a single grain of6.5% by weight noted). Despite persistent minor Mg, cal-cium was rarely observed with SEM. Low potassiumabundances are also common (3.5-8.0% by weight). Boththe excavated and source white pigments contain elevatedtitanium-enriched zones consistent with the elevatedtitanium levels observed in some of the EDS spectra(�25-50% by weight; the lighter sections shown in theSEM backscattered micrographs of Figure 23 are titan-ium-rich). Small grains, bright under the backscatterdetector, were shown with EDS to contain one dominantmetal, either Ti, Fe, Cu or Zn (the sub-micron whitegrains in Figure 23). Only one isolated feature containingCa (�17% by weight) was observed in the sample fromthe second location in the white pigment source. Thenear-absence of Ca, and occurrence of Mg as a persistentminor chemical component associated with Si and Al(not Ca), confirmed the very low Ca measured by pXRFand determined that the white pigments cannotbe huntite.
Importantly, white pigment sources (Clarke 1976;Thomas 1998) and white pigment known to have beenused to paint Wanjina motifs in the Kimberley have pre-viously been identified as mixed minerals (Ford et al.1994; Huntley et al. 2014; Ward et al. 2001; Watchman1997). The variations in clay-rich white pigmentsobserved in the excavated samples from Borologa 1 andnearby source are therefore consistent with these previouscharacterisations. The presence of P and Mg detected bySEM in the excavated Borologa 1 and nearby source sam-ples likely reflects minor mica mineral phases such asphlogopite (KMg3AlSi3O10F(OH)), which had previouslybeen identified in Kimberley rock art in the LowerMitchell Falls area c. 135 km west-northwest of Borologa1 (Huntley et al. 2015).
Internal variation was noted in the morphology andcomposition of the Drysdale River white pigment sourcelocations near Borologa 1. Of the subsamples from twoadjacent locations along the exposed clay bed, samplesfrom the second have higher silicon content than the firstsource location and the excavated archaeological speci-mens. The general morphology of the pigment from thesecond source location is more amorphous, likely relatedto silica enrichment (Figures 23C, 23D).
Previous mineralogical analyses at other sites in theKimberley identified minor phases of iron-rich mineralsin white Wanjina paint flakes, though this has been
AUSTRALIAN ARCHAEOLOGY 37
presumed to be due to weathering processes on the artpanel surfaces associated with their tropical environment(Ford et al. 1994; Huntley et al. 2014). Here we haveshown that Fe is common in both excavated white pig-ments at Borologa 1 and in the nearby white clay source,and that the excavated white pigment was probably local(Table 8).
The predominance of Al and Si was also measured inrelative abundances via pXRF for the excavated red ochrefragments (Table 8). Combined with the absence of Ca,again we interpret this as evidence of a clay matrix forthe red ochres, indicating a similar local geomorphicenvironment for its source(s) (Huntley 2015).
38 B. DAVID ETAL.